seqsero2: libs/sratoolkit.2.8.0-centos_linux64/schema/ncbi/wgs-contig.vschema annotate

author	charles_s_test
date	Mon, 27 Nov 2017 11:21:07 -0500
parents
children

rev	line source
3 38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	1 /*===========================================================================
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	2 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	3 * PUBLIC DOMAIN NOTICE
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	4 * National Center for Biotechnology Information
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	5 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	6 * This software/database is a "United States Government Work" under the
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	7 * terms of the United States Copyright Act. It was written as part of
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	8 * the author's official duties as a United States Government employee and
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	9 * thus cannot be copyrighted. This software/database is freely available
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	10 * to the public for use. The National Library of Medicine and the U.S.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	11 * Government have not placed any restriction on its use or reproduction.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	12 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	13 * Although all reasonable efforts have been taken to ensure the accuracy
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	14 * and reliability of the software and data, the NLM and the U.S.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	15 * Government do not and cannot warrant the performance or results that
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	16 * may be obtained by using this software or data. The NLM and the U.S.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	17 * Government disclaim all warranties, express or implied, including
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	18 * warranties of performance, merchantability or fitness for any particular
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	19 * purpose.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	20 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	21 * Please cite the author in any work or product based on this material.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	22 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	23 * ===========================================================================
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	24 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	25 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	26
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	27 /*==========================================================================
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	28 * WGS Contig
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	29 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	30 version 1;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	31
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	32 include 'vdb/vdb.vschema';
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	33 include 'ncbi/ncbi.vschema';
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	34 include 'ncbi/seq.vschema';
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	35 include 'ncbi/spotname.vschema';
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	36 include 'ncbi/stats.vschema';
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	37
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	38
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	39 /*--------------------------------------------------------------------------
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	40 * types
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	41 * constants
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	42 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	43
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	44 /* component_props
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	45 * a signed value describing contig or gap components of scaffolds, or
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	46 * gaps in contig sequences.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	47 * Positive values refer to contigs and negatives describe gaps
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	48 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	49 typedef I16 NCBI:WGS:component_props;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	50
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	51 /* component description
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	52 * the sequencing status of the component
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	53 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	54 * These typically correspond to keywords in the INSDC submission.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	55 * Current acceptable values are:
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	56 * A Active Finishing
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	57 * D Draft HTG (often phase1 and phase2 are called Draft,
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	58 * whether or not they have the draft keyword).
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	59 * F Finished HTG (phase3)
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	60 * G Whole Genome Finishing
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	61 * O Other sequence (typically means no HTG keyword)
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	62 * P Pre Draft
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	63 * W WGS contig
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	64 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	65 const NCBI:WGS:component_props NCBI:WGS:component:WGS = 0;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	66 const NCBI:WGS:component_props NCBI:WGS:component:ActiveFinishing = 1;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	67 const NCBI:WGS:component_props NCBI:WGS:component:DraftHTG = 2;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	68 const NCBI:WGS:component_props NCBI:WGS:component:FinishedHTG = 3;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	69 const NCBI:WGS:component_props NCBI:WGS:component:WholeGenomeFinishing = 4;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	70 const NCBI:WGS:component_props NCBI:WGS:component:OtherSequence = 5;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	71 const NCBI:WGS:component_props NCBI:WGS:component:PreDraft = 6;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	72
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	73 /* strand
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	74 * specifies the orientation of the component relative to scaffold
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	75 * values given allow strand to be determined as "prop / 16"
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	76 * yielding:
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	77 * 0 unknown orientation
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	78 * 1 plus strand
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	79 * 2 negative strand
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	80 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	81 const NCBI:WGS:component_props NCBI:WGS:strand:plus = 16;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	82 const NCBI:WGS:component_props NCBI:WGS:strand:minus = 32;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	83
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	84
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	85 /* gap description
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	86 * These typically correspond to keywords in the INSDC submission.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	87 * Current acceptable values are:
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	88 * N gap with specified size
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	89 * U gap of unknown size, defaulting to 100 bases.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	90 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	91 const NCBI:WGS:component_props NCBI:WGS:gap:known = -1;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	92 const NCBI:WGS:component_props NCBI:WGS:gap:unknown = -2;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	93
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	94 /* gap type
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	95 * scaffold a gap between two sequence contigs in a scaffold
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	96 * contig an unspanned gap between two sequence contigs
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	97 * centromere a gap inserted for the centromere
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	98 * short_arm a gap inserted at the start of an acrocentric chromosome
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	99 * heterochromatin a gap inserted for an especially large region of heterochromatic sequence
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	100 * telomere a gap inserted for the telomere
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	101 * repeat an unresolvable repeat
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	102 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	103 const NCBI:WGS:component_props NCBI:WGS:gap:scaffold = -4;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	104 const NCBI:WGS:component_props NCBI:WGS:gap:contig = -8;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	105 const NCBI:WGS:component_props NCBI:WGS:gap:centromere = -12;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	106 const NCBI:WGS:component_props NCBI:WGS:gap:short_arm = -16;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	107 const NCBI:WGS:component_props NCBI:WGS:gap:heterochromatin = -20;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	108 const NCBI:WGS:component_props NCBI:WGS:gap:telomere = -24;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	109 const NCBI:WGS:component_props NCBI:WGS:gap:repeat = -28;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	110
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	111 /* gap_linkage
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	112 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	113 typedef I32 NCBI:WGS:gap_linkage;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	114
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	115 /* gap linkage and linkage evidence
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	116 * There can be multiple linkage evidences or linkage with no evidence
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	117 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	118 * paired-ends paired sequences from the two ends of a DNA fragment
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	119 * align_genus alignment to a reference genome within the same genus
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	120 * align_xgenus alignment to a reference genome within another genus
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	121 * align_trnscpt alignment to a transcript from the same species
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	122 * within_clone sequence on both sides of the gap is derived from
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	123 * the same clone, but the gap is not spanned by paired-ends
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	124 * clone_contig linkage is provided by a clone contig in the tiling path
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	125 * map linkage asserted using a non-sequence based map
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	126 * such as RH, linkage, fingerprint or optical
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	127 * strobe strobe sequencing (PacBio)
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	128 * unspecified
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	129 * pcr PCR
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	130 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	131 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage:linked = 1;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	132 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:paired_ends = 2;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	133 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:align_genus = 4;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	134 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:align_xgenus = 8;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	135 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:align_trnscpt = 16;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	136 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:within_clone = 32;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	137 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:clone_contig = 64;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	138 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:map = 128;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	139 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:strobe = 256;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	140 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:unspecified = 512;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	141 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:pcr = 1024;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	142
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	143 /*--------------------------------------------------------------------------
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	144 * functions
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	145 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	146
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	147 /* tokenize_nuc_accession
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	148 * tokenize_prot_accession
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	149 * scans name on input
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	150 * tokenizes into parts
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	151 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	152 extern function text:token
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	153 NCBI:WGS:tokenize_nuc_accession #1 ( ascii acc );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	154 extern function text:token
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	155 NCBI:WGS:tokenize_prot_accession #1 ( ascii acc );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	156
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	157 const U16 NCBI:WGS:acc_token:unrecognized = 1;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	158 const U16 NCBI:WGS:acc_token:prefix = 2;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	159 const U16 NCBI:WGS:acc_token:contig = 3;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	160
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	161
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	162 /* build_scaffold_read
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	163 * assembles contigs and gaps into a single row
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	164 * transcribes + strand contigs as they are,
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	165 * performs reverse complement of - strand contigs,
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	166 * fills gaps with stated number of N
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	167 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	168 * build_scaffold_quality
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	169 * assembles contig and gap qualities into a single row
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	170 * contig qualities are taken as they are,
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	171 * gap qualities are assigned a constant
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	172 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	173 * "component_start" [ DATA ] - starting locations on each
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	174 * component or 0 for gaps. normal starting point is 0,
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	175 * but offsets are supported.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	176 * NB - ONE-BASED COORDINATES
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	177 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	178 * "component_len" [ DATA ] - length of contig sequence
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	179 * from component_start, or length of gap, projected onto
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	180 * scaffold at scaffold_start.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	181 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	182 * "component_props" [ DATA ] - see discussion of type
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	183 * distinguish between contigs and gaps, indicate strand
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	184 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	185 * "component_id" [ DATA ] - foreign keys into SEQUENCE table
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	186 * row_len ( component_id ) == count-of-contigs ( component_props )
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	187 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	188 extern function INSDC:4na:bin NCBI:WGS:build_scaffold_read #1
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	189 ( INSDC:coord:one component_start, INSDC:coord:len component_len,
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	190 NCBI:WGS:component_props component_props, I64 component_id );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	191
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	192 extern function INSDC:quality:phred NCBI:WGS:build_scaffold_qual #1
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	193 ( INSDC:coord:one component_start, INSDC:coord:len component_len,
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	194 NCBI:WGS:component_props component_props, I64 component_id );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	195
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	196
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	197 /* build_read_type
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	198 * generate standard SRA read type from component properties
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	199 * contigs are biological, gaps are technical
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	200 *
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	201 * "component_props" [ DATA ] - see discussion of type
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	202 * distinguish between contigs and gaps, indicate strand
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	203 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	204 extern function INSDC:SRA:xread_type
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	205 NCBI:WGS:build_read_type #1 ( NCBI:WGS:component_props component_props );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	206
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	207
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	208 /*--------------------------------------------------------------------------
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	209 * nucleotide
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	210 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	211 table NCBI:WGS:tbl:nucleotide #1.1
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	212 = NCBI:tbl:base_space #2.0.3
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	213 , NCBI:tbl:phred_quality #2.0.4
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	214 , NCBI:SRA:tbl:stats #1.2.0
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	215 {
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	216 /* ACCESSION
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	217 * [<opt-prefix>]<4-letter-prefix><2-digit-version><6-or-7-digit-contig>
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	218 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	219 extern column ascii ACCESSION = out_accession;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	220 extern column U32 ACC_VERSION = .ACC_VERSION \| <U32> echo <1> ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	221
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	222 // input
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	223 ascii in_accession = ACCESSION;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	224
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	225 // parsed input
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	226 text:token in_acc_token
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	227 = NCBI:WGS:tokenize_nuc_accession ( in_accession );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	228 // [<opt-prefix>]<4-letter-prefix><2-digit-version>
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	229 ascii in_acc_prefix
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	230 = extract_token < 0 > ( in_accession, in_acc_token );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	231 // <6-or-7-digit-contig>
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	232 ascii in_contig_text
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	233 = extract_token < 1 > ( in_accession, in_acc_token );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	234 U32 in_contig_len
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	235 = row_len ( in_contig_text );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	236 U64 in_contig_bin
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	237 = strtonum ( in_contig_text );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	238
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	239 // physical storage
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	240 physical column < ascii > zip_encoding .ACC_PREFIX = in_acc_prefix;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	241 physical column < U32 > izip_encoding .ACC_CONTIG_LEN = in_contig_len;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	242 physical column < U64 > izip_encoding .ACC_CONTIG = in_contig_bin;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	243 physical column < U32 > izip_encoding .ACC_VERSION = ACC_VERSION; //needed to back-fill WGS data from ID where version may be > 1
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	244
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	245 // output
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	246 ascii out_acc_prefix
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	247 = .ACC_PREFIX
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	248 \| < ascii > meta:read < 'ACC_PREFIX', true > ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	249 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	250 U32 out_acc_contig_len
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	251 = .ACC_CONTIG_LEN
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	252 \| < U32 > meta:value < 'ACC_CONTIG_LEN', true > ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	253 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	254 U64 out_acc_contig
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	255 = .ACC_CONTIG
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	256 \| ( U64 ) row_id ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	257 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	258 ascii out_accession
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	259 = sprintf < "%s%0*u" > ( out_acc_prefix, out_acc_contig_len, out_acc_contig );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	260
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	261 readonly column ascii ACC_PREFIX
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	262 = .ACC_PREFIX
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	263 \| < ascii > meta:read < 'ACC_PREFIX', true > ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	264 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	265 readonly column U32 ACC_CONTIG_LEN
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	266 = .ACC_CONTIG_LEN
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	267 \| < U32 > meta:value < 'ACC_CONTIG_LEN', true > ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	268 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	269
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	270 /* CONTIG_NAME
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	271 * principal name
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	272 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	273 extern column utf8 CONTIG_NAME
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	274 = idx:text:project #1.0 < 'contig_name' > ( .CONTIG_NAME );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	275
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	276 physical column < utf8 > zip_encoding .CONTIG_NAME
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	277 = idx:text:insert #1.0 < 'contig_name' > ( CONTIG_NAME );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	278
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	279 ascii out_contig_name = cast ( CONTIG_NAME );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	280
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	281 // NB - this is only useful if CONTIG_NAME is unique
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	282 // or if clustered by CONTIG_NAME
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	283 readonly column vdb:row_id_range CONTIG_NAME_ROW_RANGE
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	284 = idx:text:lookup #1.0 < 'contig_name', 'NAME_QUERY' > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	285
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	286 /* EXTRA_SEQIDS
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	287 * pipe-separated list of additional names
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	288 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	289 extern column < ascii > zip_encoding EXTRA_SEQIDS;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	290
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	291 /* TITLE
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	292 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	293 extern column < ascii > zip_encoding TITLE;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	294
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	295 /* GI
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	296 * gi is indexed in a parallel table
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	297 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	298 extern column < NCBI:gi > izip_encoding GI;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	299
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	300 /* TAXID
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	301 * taxonomy id
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	302 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	303 extern column < NCBI:taxid > izip_encoding TAXID;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	304
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	305 /* GB_STATE
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	306 * genbank state
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	307 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	308 extern column < NCBI:gb_state > izip_encoding GB_STATE;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	309
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	310 /* DESCR
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	311 * ASN.1 description
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	312 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	313 extern column < NCBI:asn:binary > zip_encoding DESCR;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	314
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	315 /* ANNOT
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	316 * ASN.1 annotation
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	317 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	318 extern column < NCBI:asn:binary > zip_encoding ANNOT;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	319
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	320 /* GAP_START
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	321 * Starting position of a gap
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	322 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	323 extern column < INSDC:coord:zero > izip_encoding GAP_START;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	324
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	325 /* GAP_LEN
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	326 * Length of a gap
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	327 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	328 extern column < INSDC:coord:len > izip_encoding GAP_LEN;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	329
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	330 /* GAP_PROPS
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	331 * See description of type
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	332 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	333 extern column < NCBI:WGS:component_props > zip_encoding GAP_PROPS;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	334
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	335 /* GAP_LINKAGE
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	336 * See description of type
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	337 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	338 extern column < NCBI:WGS:gap_linkage> zip_encoding GAP_LINKAGE;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	339
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	340
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	341 ascii out_seqid_gi = sprintf < "gi\|%u" > ( .GI );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	342 ascii out_seqid_gb = sprintf < "gb\|%s.%u\|" > ( out_accession, ACC_VERSION )
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	343 \| sprintf < "gb\|%s.1\|" > ( out_accession );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	344 ascii out_seqid_gnl = sprintf < "gnl\|WGS:%s\|%s" > (.ACC_PREFIX, out_contig_name ) \| <ascii> echo < '' > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	345
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	346
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	347 /* outputs to spotname */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	348 ascii out_seqid_name
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	349 = sprintf < "%s\|%s" > ( out_seqid_gi , out_seqid_gb )
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	350 \| sprintf < "%s" > ( out_seqid_gb );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	351
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	352 ascii out_spot_name = sprintf < "%s %s" > ( out_seqid_name, .TITLE );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	353
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	354 readonly column ascii SEQ_ID = out_seqid_name;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	355 readonly column ascii SEQ_ID_GNL = out_seqid_gnl;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	356
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	357 /* outputs to spotdesc */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	358 // INSDC:coord:len in_read_len = (INSDC:coord:len) row_len ( in_2na_bin );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	359 INSDC:coord:len out_read_len = (INSDC:coord:len) row_len ( out_2na_bin );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	360 INSDC:coord:len trim_len = (INSDC:coord:len) row_len ( out_2na_bin );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	361 INSDC:coord:zero out_read_start = <INSDC:coord:zero> echo < 0 > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	362 INSDC:coord:zero trim_start = <INSDC:coord:zero> echo < 0 > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	363 INSDC:SRA:read_filter out_rd_filter = < INSDC:SRA:read_filter > echo < SRA_READ_FILTER_PASS > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	364 INSDC:SRA:xread_type out_read_type = < INSDC:SRA:xread_type > echo < SRA_READ_TYPE_BIOLOGICAL > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	365 // help trigger statistics
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	366 INSDC:SRA:xread_type _alt_in_read_type = < INSDC:SRA:xread_type > echo < SRA_READ_TYPE_BIOLOGICAL > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	367 INSDC:coord:len _alt_in_read_len = (INSDC:coord:len) row_len ( in_2na_bin );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	368
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	369 ascii out_label = < ascii > echo < "contig" > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	370 INSDC:coord:len out_label_len = < INSDC:coord:len > echo < 6 > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	371 INSDC:coord:zero out_label_start = < INSDC:coord:zero > echo < 0 > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	372
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	373 INSDC:SRA:platform_id out_platform = < INSDC:SRA:platform_id > echo < SRA_PLATFORM_UNDEFINED > ();
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	374 };
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	375
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	376
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	377 /*--------------------------------------------------------------------------
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	378 * protein
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	379 * contig
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	380 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	381 table NCBI:WGS:tbl:protein #1
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	382 = NCBI:tbl:protein #1.0.0
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	383 {
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	384 /* ACCESSION
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	385 * [<opt-prefix>]<4-letter-prefix><2-digit-version><6-or-7-digit-contig>
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	386 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	387 extern column ascii ACCESSION = out_accession;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	388
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	389 // input
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	390 ascii in_accession = ACCESSION;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	391
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	392 // parsed input
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	393 text:token in_acc_token
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	394 = NCBI:WGS:tokenize_prot_accession ( in_accession );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	395 // [<opt-prefix>]<4-letter-prefix><2-digit-version>
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	396 ascii in_acc_prefix
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	397 = extract_token < 0 > ( in_accession, in_acc_token );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	398 // <6-or-7-digit-contig>
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	399 ascii in_contig_text
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	400 = extract_token < 1 > ( in_accession, in_acc_token );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	401 U32 in_contig_len
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	402 = row_len ( in_contig_text );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	403 I64 in_contig_bin
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	404 = strtonum ( in_contig_text );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	405
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	406 // physical storage
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	407 physical column < ascii > zip_encoding .ACC_PREFIX = in_acc_prefix;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	408 physical column < U32 > izip_encoding .ACC_CONTIG_LEN = in_contig_len;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	409 physical column < U64 > izip_encoding .ACC_CONTIG = in_contig_bin;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	410
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	411 // output
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	412 ascii out_acc_prefix
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	413 = .ACC_PREFIX
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	414 \| < ascii > meta:read < 'ACC_PREFIX', true > ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	415 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	416 U32 out_acc_contig_len
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	417 = .ACC_CONTIG_LEN
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	418 \| < U32 > meta:value < 'ACC_CONTIG_LEN', true > ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	419 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	420 U64 out_acc_contig
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	421 = .ACC_CONTIG
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	422 \| ( U64 ) row_id ()
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	423 ;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	424 ascii out_accession
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	425 = sprintf < "%s%0*u" > ( out_acc_prefix, out_acc_contig_len, out_acc_contig );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	426
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	427 /* TITLE
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	428 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	429 extern column < ascii > zip_encoding TITLE;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	430
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	431 /* GI
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	432 * gi is indexed in a parallel table
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	433 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	434 extern column < NCBI:gi > izip_encoding GI;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	435
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	436 /* GB_STATE
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	437 * genbank state
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	438 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	439 extern column < NCBI:gb_state > izip_encoding GB_STATE;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	440
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	441 /* DESCR
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	442 * ASN.1 description
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	443 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	444 extern column < NCBI:asn:binary > zip_encoding DESCR;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	445
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	446 /* ANNOT
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	447 * ASN.1 annotation
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	448 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	449 extern column < NCBI:asn:binary > zip_encoding ANNOT;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	450
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	451 /* outputs to spotname */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	452 ascii out_seqid_name = sprintf < "TBD" > ( .ACC_PREFIX, out_accession );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	453 ascii out_spot_name = sprintf < "%s %s" > ( out_seqid_name, .TITLE );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	454
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	455 /* TBD
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	456 * need to create an extension to NCBI:tbl:protein
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	457 * that satisfies fastq-dump requirements for READ and QUALITY
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	458 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	459 };
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	460
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	461
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	462 /*--------------------------------------------------------------------------
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	463 * gi_idx
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	464 * gi is row-id
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	465 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	466 table NCBI:WGS:tbl:gi_idx #1
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	467 {
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	468 /* NUC_ROW_ID
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	469 * row-id in nucleotide table
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	470 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	471 extern column < I64 > izip_encoding NUC_ROW_ID;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	472
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	473 /* PROT_ROW_ID
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	474 * row-id in protein table
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	475 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	476 extern column < I64 > izip_encoding PROT_ROW_ID;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	477 };
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	478
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	479
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	480 /*--------------------------------------------------------------------------
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	481 * scaffold
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	482 * records AGP data
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	483 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	484 table NCBI:WGS:tbl:scaffold #1
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	485 {
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	486 /* SCAFFOLD_NAME
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	487 * This is the identifier for the object being assembled.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	488 * This can be a chromosome, scaffold or contig.
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	489 * If an accession.version identifier is not used to describe
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	490 * the object the naming convention is to precede chromosome numbers
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	491 * (e.g. chr1) and linkage group numbers (e.g. LG3).
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	492 * Contigs or scaffolds may have any identifier that is unique
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	493 * within the assembly
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	494 */
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	495 extern column utf8 SCAFFOLD_NAME
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	496 = out_scaffold_name;
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	497 extern column ascii SCAFFOLD_NAME = cast (out_scaffold_name);
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	498 utf8 out_scaffold_name
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	499 = idx:text:project #1.0 < 'scaffold_name' > ( .SCAFFOLD_NAME );
38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty charles_s_test parents: diff changeset	500 physical column < utf8 > zip_encoding .SCAFFOLD_NAME

3

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

1 /*===========================================================================

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

2 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

3 * PUBLIC DOMAIN NOTICE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

4 * National Center for Biotechnology Information

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

5 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

6 * This software/database is a "United States Government Work" under the

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

7 * terms of the United States Copyright Act. It was written as part of

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

8 * the author's official duties as a United States Government employee and

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

9 * thus cannot be copyrighted. This software/database is freely available

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

10 * to the public for use. The National Library of Medicine and the U.S.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

11 * Government have not placed any restriction on its use or reproduction.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

12 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

13 * Although all reasonable efforts have been taken to ensure the accuracy

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

14 * and reliability of the software and data, the NLM and the U.S.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

15 * Government do not and cannot warrant the performance or results that

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

16 * may be obtained by using this software or data. The NLM and the U.S.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

17 * Government disclaim all warranties, express or implied, including

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

18 * warranties of performance, merchantability or fitness for any particular

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

19 * purpose.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

20 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

21 * Please cite the author in any work or product based on this material.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

22 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

23 * ===========================================================================

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

24 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

25 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

26

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

27 /*==========================================================================

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

28 * WGS Contig

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

29 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

30 version 1;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

31

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

32 include 'vdb/vdb.vschema';

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

33 include 'ncbi/ncbi.vschema';

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

34 include 'ncbi/seq.vschema';

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

35 include 'ncbi/spotname.vschema';

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

36 include 'ncbi/stats.vschema';

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

37

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

38

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

39 /*--------------------------------------------------------------------------

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

40 * types

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

41 * constants

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

42 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

43

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

44 /* component_props

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

45 * a signed value describing contig or gap components of scaffolds, or

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

46 * gaps in contig sequences.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

47 * Positive values refer to contigs and negatives describe gaps

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

48 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

49 typedef I16 NCBI:WGS:component_props;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

50

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

51 /* component description

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

52 * the sequencing status of the component

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

53 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

54 * These typically correspond to keywords in the INSDC submission.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

55 * Current acceptable values are:

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

56 * A Active Finishing

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

57 * D Draft HTG (often phase1 and phase2 are called Draft,

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

58 * whether or not they have the draft keyword).

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

59 * F Finished HTG (phase3)

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

60 * G Whole Genome Finishing

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

61 * O Other sequence (typically means no HTG keyword)

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

62 * P Pre Draft

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

63 * W WGS contig

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

64 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

65 const NCBI:WGS:component_props NCBI:WGS:component:WGS = 0;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

66 const NCBI:WGS:component_props NCBI:WGS:component:ActiveFinishing = 1;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

67 const NCBI:WGS:component_props NCBI:WGS:component:DraftHTG = 2;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

68 const NCBI:WGS:component_props NCBI:WGS:component:FinishedHTG = 3;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

69 const NCBI:WGS:component_props NCBI:WGS:component:WholeGenomeFinishing = 4;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

70 const NCBI:WGS:component_props NCBI:WGS:component:OtherSequence = 5;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

71 const NCBI:WGS:component_props NCBI:WGS:component:PreDraft = 6;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

72

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

73 /* strand

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

74 * specifies the orientation of the component relative to scaffold

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

75 * values given allow strand to be determined as "prop / 16"

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

76 * yielding:

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

77 * 0 unknown orientation

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

78 * 1 plus strand

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

79 * 2 negative strand

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

80 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

81 const NCBI:WGS:component_props NCBI:WGS:strand:plus = 16;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

82 const NCBI:WGS:component_props NCBI:WGS:strand:minus = 32;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

83

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

84

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

85 /* gap description

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

86 * These typically correspond to keywords in the INSDC submission.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

87 * Current acceptable values are:

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

88 * N gap with specified size

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

89 * U gap of unknown size, defaulting to 100 bases.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

90 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

91 const NCBI:WGS:component_props NCBI:WGS:gap:known = -1;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

92 const NCBI:WGS:component_props NCBI:WGS:gap:unknown = -2;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

93

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

94 /* gap type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

95 * scaffold a gap between two sequence contigs in a scaffold

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

96 * contig an unspanned gap between two sequence contigs

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

97 * centromere a gap inserted for the centromere

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

98 * short_arm a gap inserted at the start of an acrocentric chromosome

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

99 * heterochromatin a gap inserted for an especially large region of heterochromatic sequence

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

100 * telomere a gap inserted for the telomere

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

101 * repeat an unresolvable repeat

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

102 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

103 const NCBI:WGS:component_props NCBI:WGS:gap:scaffold = -4;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

104 const NCBI:WGS:component_props NCBI:WGS:gap:contig = -8;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

105 const NCBI:WGS:component_props NCBI:WGS:gap:centromere = -12;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

106 const NCBI:WGS:component_props NCBI:WGS:gap:short_arm = -16;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

107 const NCBI:WGS:component_props NCBI:WGS:gap:heterochromatin = -20;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

108 const NCBI:WGS:component_props NCBI:WGS:gap:telomere = -24;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

109 const NCBI:WGS:component_props NCBI:WGS:gap:repeat = -28;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

110

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

111 /* gap_linkage

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

112 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

113 typedef I32 NCBI:WGS:gap_linkage;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

114

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

115 /* gap linkage and linkage evidence

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

116 * There can be multiple linkage evidences or linkage with no evidence

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

117 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

118 * paired-ends paired sequences from the two ends of a DNA fragment

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

119 * align_genus alignment to a reference genome within the same genus

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

120 * align_xgenus alignment to a reference genome within another genus

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

121 * align_trnscpt alignment to a transcript from the same species

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

122 * within_clone sequence on both sides of the gap is derived from

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

123 * the same clone, but the gap is not spanned by paired-ends

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

124 * clone_contig linkage is provided by a clone contig in the tiling path

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

125 * map linkage asserted using a non-sequence based map

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

126 * such as RH, linkage, fingerprint or optical

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

127 * strobe strobe sequencing (PacBio)

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

128 * unspecified

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

129 * pcr PCR

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

130 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

131 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage:linked = 1;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

132 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:paired_ends = 2;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

133 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:align_genus = 4;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

134 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:align_xgenus = 8;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

135 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:align_trnscpt = 16;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

136 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:within_clone = 32;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

137 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:clone_contig = 64;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

138 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:map = 128;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

139 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:strobe = 256;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

140 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:unspecified = 512;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

141 const NCBI:WGS:gap_linkage NCBI:WGS:gap:linkage_evidence:pcr = 1024;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

142

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

143 /*--------------------------------------------------------------------------

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

144 * functions

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

145 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

146

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

147 /* tokenize_nuc_accession

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

148 * tokenize_prot_accession

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

149 * scans name on input

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

150 * tokenizes into parts

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

151 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

152 extern function text:token

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

153 NCBI:WGS:tokenize_nuc_accession #1 ( ascii acc );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

154 extern function text:token

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

155 NCBI:WGS:tokenize_prot_accession #1 ( ascii acc );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

156

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

157 const U16 NCBI:WGS:acc_token:unrecognized = 1;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

158 const U16 NCBI:WGS:acc_token:prefix = 2;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

159 const U16 NCBI:WGS:acc_token:contig = 3;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

160

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

161

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

162 /* build_scaffold_read

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

163 * assembles contigs and gaps into a single row

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

164 * transcribes + strand contigs as they are,

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

165 * performs reverse complement of - strand contigs,

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

166 * fills gaps with stated number of N

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

167 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

168 * build_scaffold_quality

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

169 * assembles contig and gap qualities into a single row

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

170 * contig qualities are taken as they are,

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

171 * gap qualities are assigned a constant

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

172 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

173 * "component_start" [ DATA ] - starting locations on each

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

174 * component or 0 for gaps. normal starting point is 0,

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

175 * but offsets are supported.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

176 * NB - ONE-BASED COORDINATES

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

177 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

178 * "component_len" [ DATA ] - length of contig sequence

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

179 * from component_start, or length of gap, projected onto

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

180 * scaffold at scaffold_start.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

181 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

182 * "component_props" [ DATA ] - see discussion of type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

183 * distinguish between contigs and gaps, indicate strand

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

184 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

185 * "component_id" [ DATA ] - foreign keys into SEQUENCE table

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

186 * row_len ( component_id ) == count-of-contigs ( component_props )

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

187 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

188 extern function INSDC:4na:bin NCBI:WGS:build_scaffold_read #1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

189 ( INSDC:coord:one component_start, INSDC:coord:len component_len,

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

190 NCBI:WGS:component_props component_props, I64 component_id );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

191

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

192 extern function INSDC:quality:phred NCBI:WGS:build_scaffold_qual #1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

193 ( INSDC:coord:one component_start, INSDC:coord:len component_len,

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

194 NCBI:WGS:component_props component_props, I64 component_id );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

195

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

196

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

197 /* build_read_type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

198 * generate standard SRA read type from component properties

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

199 * contigs are biological, gaps are technical

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

200 *

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

201 * "component_props" [ DATA ] - see discussion of type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

202 * distinguish between contigs and gaps, indicate strand

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

203 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

204 extern function INSDC:SRA:xread_type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

205 NCBI:WGS:build_read_type #1 ( NCBI:WGS:component_props component_props );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

206

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

207

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

208 /*--------------------------------------------------------------------------

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

209 * nucleotide

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

210 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

211 table NCBI:WGS:tbl:nucleotide #1.1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

212 = NCBI:tbl:base_space #2.0.3

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

213 , NCBI:tbl:phred_quality #2.0.4

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

214 , NCBI:SRA:tbl:stats #1.2.0

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

215 {

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

216 /* ACCESSION

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

217 * [<opt-prefix>]<4-letter-prefix><2-digit-version><6-or-7-digit-contig>

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

218 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

219 extern column ascii ACCESSION = out_accession;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

220 extern column U32 ACC_VERSION = .ACC_VERSION | <U32> echo <1> ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

221

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

222 // input

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

223 ascii in_accession = ACCESSION;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

224

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

225 // parsed input

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

226 text:token in_acc_token

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

227 = NCBI:WGS:tokenize_nuc_accession ( in_accession );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

228 // [<opt-prefix>]<4-letter-prefix><2-digit-version>

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

229 ascii in_acc_prefix

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

230 = extract_token < 0 > ( in_accession, in_acc_token );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

231 // <6-or-7-digit-contig>

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

232 ascii in_contig_text

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

233 = extract_token < 1 > ( in_accession, in_acc_token );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

234 U32 in_contig_len

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

235 = row_len ( in_contig_text );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

236 U64 in_contig_bin

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

237 = strtonum ( in_contig_text );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

238

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

239 // physical storage

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

240 physical column < ascii > zip_encoding .ACC_PREFIX = in_acc_prefix;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

241 physical column < U32 > izip_encoding .ACC_CONTIG_LEN = in_contig_len;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

242 physical column < U64 > izip_encoding .ACC_CONTIG = in_contig_bin;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

243 physical column < U32 > izip_encoding .ACC_VERSION = ACC_VERSION; //needed to back-fill WGS data from ID where version may be > 1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

244

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

245 // output

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

246 ascii out_acc_prefix

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

247 = .ACC_PREFIX

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

248 | < ascii > meta:read < 'ACC_PREFIX', true > ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

249 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

250 U32 out_acc_contig_len

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

251 = .ACC_CONTIG_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

252 | < U32 > meta:value < 'ACC_CONTIG_LEN', true > ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

253 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

254 U64 out_acc_contig

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

255 = .ACC_CONTIG

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

256 | ( U64 ) row_id ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

257 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

258 ascii out_accession

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

259 = sprintf < "%s%0*u" > ( out_acc_prefix, out_acc_contig_len, out_acc_contig );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

260

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

261 readonly column ascii ACC_PREFIX

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

262 = .ACC_PREFIX

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

263 | < ascii > meta:read < 'ACC_PREFIX', true > ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

264 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

265 readonly column U32 ACC_CONTIG_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

266 = .ACC_CONTIG_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

267 | < U32 > meta:value < 'ACC_CONTIG_LEN', true > ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

268 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

269

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

270 /* CONTIG_NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

271 * principal name

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

272 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

273 extern column utf8 CONTIG_NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

274 = idx:text:project #1.0 < 'contig_name' > ( .CONTIG_NAME );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

275

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

276 physical column < utf8 > zip_encoding .CONTIG_NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

277 = idx:text:insert #1.0 < 'contig_name' > ( CONTIG_NAME );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

278

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

279 ascii out_contig_name = cast ( CONTIG_NAME );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

280

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

281 // NB - this is only useful if CONTIG_NAME is unique

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

282 // or if clustered by CONTIG_NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

283 readonly column vdb:row_id_range CONTIG_NAME_ROW_RANGE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

284 = idx:text:lookup #1.0 < 'contig_name', 'NAME_QUERY' > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

285

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

286 /* EXTRA_SEQIDS

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

287 * pipe-separated list of additional names

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

288 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

289 extern column < ascii > zip_encoding EXTRA_SEQIDS;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

290

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

291 /* TITLE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

292 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

293 extern column < ascii > zip_encoding TITLE;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

294

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

295 /* GI

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

296 * gi is indexed in a parallel table

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

297 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

298 extern column < NCBI:gi > izip_encoding GI;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

299

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

300 /* TAXID

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

301 * taxonomy id

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

302 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

303 extern column < NCBI:taxid > izip_encoding TAXID;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

304

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

305 /* GB_STATE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

306 * genbank state

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

307 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

308 extern column < NCBI:gb_state > izip_encoding GB_STATE;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

309

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

310 /* DESCR

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

311 * ASN.1 description

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

312 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

313 extern column < NCBI:asn:binary > zip_encoding DESCR;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

314

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

315 /* ANNOT

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

316 * ASN.1 annotation

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

317 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

318 extern column < NCBI:asn:binary > zip_encoding ANNOT;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

319

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

320 /* GAP_START

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

321 * Starting position of a gap

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

322 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

323 extern column < INSDC:coord:zero > izip_encoding GAP_START;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

324

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

325 /* GAP_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

326 * Length of a gap

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

327 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

328 extern column < INSDC:coord:len > izip_encoding GAP_LEN;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

329

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

330 /* GAP_PROPS

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

331 * See description of type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

332 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

333 extern column < NCBI:WGS:component_props > zip_encoding GAP_PROPS;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

334

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

335 /* GAP_LINKAGE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

336 * See description of type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

337 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

338 extern column < NCBI:WGS:gap_linkage> zip_encoding GAP_LINKAGE;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

339

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

340

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

341 ascii out_seqid_gi = sprintf < "gi|%u" > ( .GI );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

342 ascii out_seqid_gb = sprintf < "gb|%s.%u|" > ( out_accession, ACC_VERSION )

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

343 | sprintf < "gb|%s.1|" > ( out_accession );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

344 ascii out_seqid_gnl = sprintf < "gnl|WGS:%s|%s" > (.ACC_PREFIX, out_contig_name ) | <ascii> echo < '' > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

345

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

346

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

347 /* outputs to spotname */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

348 ascii out_seqid_name

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

349 = sprintf < "%s|%s" > ( out_seqid_gi , out_seqid_gb )

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

350 | sprintf < "%s" > ( out_seqid_gb );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

351

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

352 ascii out_spot_name = sprintf < "%s %s" > ( out_seqid_name, .TITLE );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

353

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

354 readonly column ascii SEQ_ID = out_seqid_name;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

355 readonly column ascii SEQ_ID_GNL = out_seqid_gnl;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

356

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

357 /* outputs to spotdesc */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

358 // INSDC:coord:len in_read_len = (INSDC:coord:len) row_len ( in_2na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

359 INSDC:coord:len out_read_len = (INSDC:coord:len) row_len ( out_2na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

360 INSDC:coord:len trim_len = (INSDC:coord:len) row_len ( out_2na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

361 INSDC:coord:zero out_read_start = <INSDC:coord:zero> echo < 0 > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

362 INSDC:coord:zero trim_start = <INSDC:coord:zero> echo < 0 > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

363 INSDC:SRA:read_filter out_rd_filter = < INSDC:SRA:read_filter > echo < SRA_READ_FILTER_PASS > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

364 INSDC:SRA:xread_type out_read_type = < INSDC:SRA:xread_type > echo < SRA_READ_TYPE_BIOLOGICAL > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

365 // help trigger statistics

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

366 INSDC:SRA:xread_type _alt_in_read_type = < INSDC:SRA:xread_type > echo < SRA_READ_TYPE_BIOLOGICAL > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

367 INSDC:coord:len _alt_in_read_len = (INSDC:coord:len) row_len ( in_2na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

368

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

369 ascii out_label = < ascii > echo < "contig" > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

370 INSDC:coord:len out_label_len = < INSDC:coord:len > echo < 6 > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

371 INSDC:coord:zero out_label_start = < INSDC:coord:zero > echo < 0 > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

372

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

373 INSDC:SRA:platform_id out_platform = < INSDC:SRA:platform_id > echo < SRA_PLATFORM_UNDEFINED > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

374 };

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

375

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

376

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

377 /*--------------------------------------------------------------------------

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

378 * protein

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

379 * contig

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

380 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

381 table NCBI:WGS:tbl:protein #1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

382 = NCBI:tbl:protein #1.0.0

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

383 {

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

384 /* ACCESSION

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

385 * [<opt-prefix>]<4-letter-prefix><2-digit-version><6-or-7-digit-contig>

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

386 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

387 extern column ascii ACCESSION = out_accession;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

388

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

389 // input

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

390 ascii in_accession = ACCESSION;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

391

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

392 // parsed input

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

393 text:token in_acc_token

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

394 = NCBI:WGS:tokenize_prot_accession ( in_accession );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

395 // [<opt-prefix>]<4-letter-prefix><2-digit-version>

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

396 ascii in_acc_prefix

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

397 = extract_token < 0 > ( in_accession, in_acc_token );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

398 // <6-or-7-digit-contig>

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

399 ascii in_contig_text

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

400 = extract_token < 1 > ( in_accession, in_acc_token );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

401 U32 in_contig_len

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

402 = row_len ( in_contig_text );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

403 I64 in_contig_bin

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

404 = strtonum ( in_contig_text );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

405

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

406 // physical storage

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

407 physical column < ascii > zip_encoding .ACC_PREFIX = in_acc_prefix;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

408 physical column < U32 > izip_encoding .ACC_CONTIG_LEN = in_contig_len;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

409 physical column < U64 > izip_encoding .ACC_CONTIG = in_contig_bin;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

410

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

411 // output

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

412 ascii out_acc_prefix

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

413 = .ACC_PREFIX

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

414 | < ascii > meta:read < 'ACC_PREFIX', true > ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

415 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

416 U32 out_acc_contig_len

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

417 = .ACC_CONTIG_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

418 | < U32 > meta:value < 'ACC_CONTIG_LEN', true > ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

419 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

420 U64 out_acc_contig

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

421 = .ACC_CONTIG

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

422 | ( U64 ) row_id ()

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

423 ;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

424 ascii out_accession

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

425 = sprintf < "%s%0*u" > ( out_acc_prefix, out_acc_contig_len, out_acc_contig );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

426

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

427 /* TITLE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

428 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

429 extern column < ascii > zip_encoding TITLE;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

430

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

431 /* GI

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

432 * gi is indexed in a parallel table

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

433 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

434 extern column < NCBI:gi > izip_encoding GI;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

435

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

436 /* GB_STATE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

437 * genbank state

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

438 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

439 extern column < NCBI:gb_state > izip_encoding GB_STATE;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

440

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

441 /* DESCR

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

442 * ASN.1 description

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

443 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

444 extern column < NCBI:asn:binary > zip_encoding DESCR;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

445

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

446 /* ANNOT

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

447 * ASN.1 annotation

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

448 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

449 extern column < NCBI:asn:binary > zip_encoding ANNOT;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

450

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

451 /* outputs to spotname */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

452 ascii out_seqid_name = sprintf < "TBD" > ( .ACC_PREFIX, out_accession );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

453 ascii out_spot_name = sprintf < "%s %s" > ( out_seqid_name, .TITLE );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

454

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

455 /* TBD

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

456 * need to create an extension to NCBI:tbl:protein

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

457 * that satisfies fastq-dump requirements for READ and QUALITY

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

458 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

459 };

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

460

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

461

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

462 /*--------------------------------------------------------------------------

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

463 * gi_idx

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

464 * gi is row-id

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

465 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

466 table NCBI:WGS:tbl:gi_idx #1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

467 {

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

468 /* NUC_ROW_ID

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

469 * row-id in nucleotide table

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

470 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

471 extern column < I64 > izip_encoding NUC_ROW_ID;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

472

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

473 /* PROT_ROW_ID

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

474 * row-id in protein table

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

475 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

476 extern column < I64 > izip_encoding PROT_ROW_ID;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

477 };

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

478

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

479

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

480 /*--------------------------------------------------------------------------

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

481 * scaffold

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

482 * records AGP data

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

483 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

484 table NCBI:WGS:tbl:scaffold #1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

485 {

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

486 /* SCAFFOLD_NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

487 * This is the identifier for the object being assembled.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

488 * This can be a chromosome, scaffold or contig.

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

489 * If an accession.version identifier is not used to describe

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

490 * the object the naming convention is to precede chromosome numbers

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

491 * (e.g. chr1) and linkage group numbers (e.g. LG3).

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

492 * Contigs or scaffolds may have any identifier that is unique

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

493 * within the assembly

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

494 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

495 extern column utf8 SCAFFOLD_NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

496 = out_scaffold_name;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

497 extern column ascii SCAFFOLD_NAME = cast (out_scaffold_name);

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

498 utf8 out_scaffold_name

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

499 = idx:text:project #1.0 < 'scaffold_name' > ( .SCAFFOLD_NAME );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

500 physical column < utf8 > zip_encoding .SCAFFOLD_NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

501 = idx:text:insert #1.0 < 'scaffold_name' > ( SCAFFOLD_NAME );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

502

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

503 /* COMPONENT_START

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

504 * starting position within the component sequence

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

505 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

506 extern column < INSDC:coord:one > izip_encoding COMPONENT_START;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

507

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

508 /* COMPONENT_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

509 * length of the component/gap projected onto the scaffold

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

510 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

511 extern column < INSDC:coord:len > izip_encoding COMPONENT_LEN;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

512

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

513 /* COMPONENT_PROPS

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

514 * see description of type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

515 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

516 extern column < NCBI:WGS:component_props > zip_encoding COMPONENT_PROPS;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

517

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

518 /* COMPONENT_ID

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

519 * one row-id for each non-gap component

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

520 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

521 extern column < I64 > izip_encoding COMPONENT_ID;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

522

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

523 /* COMPONENT_LINKAGE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

524 * see description of type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

525 * one row-id for each gap component

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

526 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

527 extern column < NCBI:WGS:gap_linkage > zip_encoding COMPONENT_LINKAGE;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

528 }

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

529

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

530 table NCBI:WGS:view:scaffold #1 = NCBI:WGS:tbl:scaffold #1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

531 {

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

532 /* ACCESSION

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

533 * scaffold accession

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

534 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

535 readonly column ascii ACCESSION

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

536 = out_accession;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

537 I64 scaffold_row_id

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

538 = row_id ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

539 I64 acc_row_id

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

540 = < I64 > echo < 1 > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

541 ascii acc_prefix

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

542 = < ascii > simple_sub_select < 'SEQUENCE', 'ACC_PREFIX' > ( acc_row_id );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

543 U32 acc_contig_len

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

544 = < U32 > simple_sub_select < 'SEQUENCE', 'ACC_CONTIG_LEN' > ( acc_row_id );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

545 ascii out_accession

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

546 = sprintf < "%sS%0*d" > ( acc_prefix, acc_contig_len, scaffold_row_id );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

547

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

548 /* READ

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

549 * base space construction of entire scaffold

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

550 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

551

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

552 // construct the read from contigs and gaps

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

553 INSDC:4na:bin out_4na_bin = NCBI:WGS:build_scaffold_read

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

554 ( .COMPONENT_START, .COMPONENT_LEN, .COMPONENT_PROPS, .COMPONENT_ID );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

555

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

556 // various READ columns

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

557 default readonly column INSDC:dna:text READ

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

558 = < INSDC:4na:bin, INSDC:dna:text > map < INSDC:4na:map:BINSET, INSDC:4na:map:CHARSET > ( out_4na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

559 readonly column INSDC:4na:bin READ

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

560 = out_4na_bin;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

561 readonly column INSDC:4na:packed READ

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

562 = pack ( out_4na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

563 readonly column INSDC:x2na:bin READ

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

564 = out_x2na_bin;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

565 INSDC:x2na:bin out_x2na_bin

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

566 = < INSDC:4na:bin, INSDC:x2na:bin > map < INSDC:4na:map:BINSET, [ 4,0,1,4,2,4,4,4,3,4,4,4,4,4,4,4 ] > ( out_4na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

567 readonly column INSDC:2na:bin READ

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

568 = out_2na_bin;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

569 INSDC:2na:bin out_2na_bin

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

570 = < INSDC:x2na:bin, INSDC:2na:bin > map < INSDC:x2na:map:BINSET, [ 0, 1, 2, 3, 0 ] > ( out_x2na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

571 readonly column INSDC:2na:packed READ

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

572 = pack ( out_2na_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

573

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

574

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

575 /* CSREAD

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

576 * base space converted to color space

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

577 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

578 default readonly column INSDC:color:text CSREAD

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

579 = < INSDC:x2cs:bin, INSDC:color:text > map < INSDC:x2cs:map:BINSET, INSDC:x2cs:map:CHARSET > ( out_x2cs_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

580 readonly column INSDC:x2cs:bin CSREAD

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

581 = out_x2cs_bin;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

582 INSDC:x2cs:bin out_x2cs_bin

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

583 = NCBI:color_from_dna ( out_x2na_bin, out_read_start, .COMPONENT_LEN, out_cs_key, out_color_matrix );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

584 readonly column INSDC:2cs:bin CSREAD

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

585 = out_2cs_bin;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

586 INSDC:2cs:bin out_2cs_bin

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

587 = < INSDC:x2cs:bin, INSDC:2cs:bin > map < INSDC:x2cs:map:BINSET, [ 0, 1, 2, 3, 0 ] > ( out_x2cs_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

588 readonly column INSDC:2cs:packed CSREAD

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

589 = pack ( out_2cs_bin );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

590

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

591 /* CS_NATIVE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

592 * is color-space the native sequence space

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

593 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

594 readonly column bool CS_NATIVE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

595 = < bool > echo < false > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

596

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

597 /* CS_KEY

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

598 * leading call given in base-space

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

599 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

600 readonly column INSDC:dna:text CS_KEY

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

601 = out_cs_key;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

602 INSDC:dna:text out_cs_key

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

603 = < INSDC:dna:text > echo < 'T' > ( .COMPONENT_LEN );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

604

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

605 /* COLOR_MATRIX

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

606 * matrix used for color-space conversions

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

607 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

608 readonly column U8 COLOR_MATRIX

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

609 = out_color_matrix;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

610 U8 out_color_matrix

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

611 = < U8 > echo < INSDC:color:default_matrix > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

612

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

613

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

614 /* QUALITY

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

615 * base or color call qualities

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

616 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

617 INSDC:quality:phred out_qual_phred = NCBI:WGS:build_scaffold_qual

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

618 ( .COMPONENT_START, .COMPONENT_LEN, .COMPONENT_PROPS, .COMPONENT_ID );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

619

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

620 // PHRED is default

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

621 default readonly column INSDC:quality:phred QUALITY

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

622 = out_qual_phred;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

623

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

624 // textual encodings

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

625 readonly column INSDC:quality:text:phred_33 QUALITY

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

626 = ( INSDC:quality:text:phred_33 ) < B8 > sum < 33 > ( out_qual_phred );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

627 readonly column INSDC:quality:text:phred_64 QUALITY

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

628 = ( INSDC:quality:text:phred_64 ) < B8 > sum < 64 > ( out_qual_phred );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

629

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

630

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

631 /* PLATFORM

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

632 * sequencing platform, if known

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

633 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

634 INSDC:SRA:platform_id out_platform

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

635 = < INSDC:SRA:platform_id > echo < SRA_PLATFORM_UNDEFINED > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

636 readonly column INSDC:SRA:platform_id PLATFORM

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

637 = out_platform;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

638

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

639 /* SPOT_ID

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

640 * support for libsra

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

641 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

642 INSDC:SRA:spotid_t out_spot_id

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

643 = cast ( scaffold_row_id );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

644 readonly column INSDC:SRA:spotid_t SPOT_ID

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

645 = out_spot_id;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

646

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

647 /* NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

648 * spot name

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

649 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

650 readonly column ascii NAME

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

651 = out_scaffold_name;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

652

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

653 /* SPOT_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

654 * TRIM_START

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

655 * TRIM_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

656 * spot descriptor

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

657 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

658 readonly column INSDC:coord:len SPOT_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

659 = out_spot_len;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

660 INSDC:coord:len out_spot_len

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

661 = < INSDC:coord:len > vec_sum ( .COMPONENT_LEN );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

662 readonly column INSDC:coord:zero TRIM_START

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

663 = < INSDC:coord:zero > echo < 0 > ();

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

664 readonly column INSDC:coord:len TRIM_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

665 = out_spot_len;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

666

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

667

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

668 /* READ_START

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

669 * READ_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

670 * READ_TYPE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

671 * read descriptor portion

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

672 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

673 readonly column INSDC:coord:zero READ_START

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

674 = out_read_start;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

675 INSDC:coord:zero out_read_start

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

676 = ( INSDC:coord:zero ) < U32 > integral ( .COMPONENT_LEN );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

677 readonly column INSDC:coord:len READ_LEN

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

678 = .COMPONENT_LEN;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

679 readonly column INSDC:SRA:xread_type READ_TYPE

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

680 = out_read_type;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

681 INSDC:SRA:xread_type out_read_type

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

682 = NCBI:WGS:build_read_type ( .COMPONENT_PROPS );

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

683 }

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

684

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

685

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

686 /*--------------------------------------------------------------------------

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

687 * contig

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

688 */

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

689 database NCBI:WGS:db:contig #1.1

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

690 {

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

691 table NCBI:WGS:tbl:nucleotide SEQUENCE;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

692 table NCBI:WGS:tbl:protein PROTEIN;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

693 table NCBI:WGS:tbl:gi_idx GI_IDX;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

694 table NCBI:WGS:view:scaffold SCAFFOLD;

38ad1130d077 planemo upload commit a4fb57231f274270afbfebd47f67df05babffa4a-dirty

charles_s_test

parents:

diff changeset

695 };

Mercurial > repos > charles_s_test > seqsero2

annotate libs/sratoolkit.2.8.0-centos_linux64/schema/ncbi/wgs-contig.vschema @ 3:38ad1130d077 draft