fastq_paired_unpaired: tools/fastq/fastq_paired

annotate tools/fastq/fastq_paired_unpaired.py @ 0:3a39d2053bc5

Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository

author	peterjc
date	Tue, 07 Jun 2011 16:32:01 -0400
parents
children	2feaef06d388

rev	line source
0 3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	1 #!/usr/bin/env python
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	2 """Divides a FASTQ into paired and single (orphan reads) as separate files.
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	3
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	4 The input file should be a valid FASTQ file which has been sorted so that
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	5 any partner forward+reverse reads are consecutive. The output files all
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	6 preserve this sort order. Pairing are recognised based on standard name
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	7 suffices. See below or run the tool with no arguments for more details.
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	8
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	9 Note that the FASTQ variant is unimportant (Sanger, Solexa, Illumina, or even
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	10 Color Space should all work equally well).
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	11
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	12 This script is copyright 2010 by Peter Cock, SCRI, UK. All rights reserved.
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	13 See accompanying text file for licence details (MIT/BSD style).
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	14
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	15 This is version 0.0.4 of the script.
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	16 """
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	17 import os
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	18 import sys
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	19 import re
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	20 from galaxy_utils.sequence.fastq import fastqReader, fastqWriter
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	21
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	22 def stop_err(msg, err=1):
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	23 sys.stderr.write(msg.rstrip() + "\n")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	24 sys.exit(err)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	25
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	26 msg = """Expect either 3 or 4 arguments, all FASTQ filenames.
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	27
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	28 If you want two output files, use four arguments:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	29 - FASTQ variant (e.g. sanger, solexa, illumina or cssanger)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	30 - Sorted input FASTQ filename,
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	31 - Output paired FASTQ filename (forward then reverse interleaved),
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	32 - Output singles FASTQ filename (orphan reads)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	33
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	34 If you want three output files, use five arguments:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	35 - FASTQ variant (e.g. sanger, solexa, illumina or cssanger)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	36 - Sorted input FASTQ filename,
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	37 - Output forward paired FASTQ filename,
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	38 - Output reverse paired FASTQ filename,
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	39 - Output singles FASTQ filename (orphan reads)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	40
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	41 The input file should be a valid FASTQ file which has been sorted so that
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	42 any partner forward+reverse reads are consecutive. The output files all
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	43 preserve this sort order.
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	44
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	45 Any reads where the forward/reverse naming suffix used is not recognised
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	46 are treated as orphan reads. The tool supports the /1 and /2 convention
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	47 used by Illumina, the .f and .r convention, and the Sanger convention
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	48 (see http://staden.sourceforge.net/manual/pregap4_unix_50.html for details).
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	49
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	50 Note that this does support multiple forward and reverse reads per template
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	51 (which is quite common with Sanger sequencing), e.g. this which is sorted
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	52 alphabetically:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	53
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	54 WTSI_1055_4p17.p1kapIBF
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	55 WTSI_1055_4p17.p1kpIBF
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	56 WTSI_1055_4p17.q1kapIBR
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	57 WTSI_1055_4p17.q1kpIBR
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	58
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	59 or this where the reads already come in pairs:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	60
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	61 WTSI_1055_4p17.p1kapIBF
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	62 WTSI_1055_4p17.q1kapIBR
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	63 WTSI_1055_4p17.p1kpIBF
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	64 WTSI_1055_4p17.q1kpIBR
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	65
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	66 both become:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	67
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	68 WTSI_1055_4p17.p1kapIBF paired with WTSI_1055_4p17.q1kapIBR
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	69 WTSI_1055_4p17.p1kpIBF paired with WTSI_1055_4p17.q1kpIBR
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	70 """
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	71
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	72 if len(sys.argv) == 5:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	73 format, input_fastq, pairs_fastq, singles_fastq = sys.argv[1:]
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	74 elif len(sys.argv) == 6:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	75 pairs_fastq = None
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	76 format, input_fastq, pairs_f_fastq, pairs_r_fastq, singles_fastq = sys.argv[1:]
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	77 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	78 stop_err(msg)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	79
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	80 format = format.replace("fastq", "").lower()
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	81 if not format:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	82 format="sanger" #safe default
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	83 elif format not in ["sanger","solexa","illumina","cssanger"]:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	84 stop_err("Unrecognised format %s" % format)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	85
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	86 def f_match(name):
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	87 if name.endswith("/1") or name.endswith(".f"):
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	88 return True
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	89
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	90 #Cope with three widely used suffix naming convensions,
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	91 #Illumina: /1 or /2
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	92 #Forward/revered: .f or .r
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	93 #Sanger, e.g. .p1k and .q1k
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	94 #See http://staden.sourceforge.net/manual/pregap4_unix_50.html
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	95 re_f = re.compile(r"(/1\|\.f\|\.[sfp]\d\w*)$")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	96 re_r = re.compile(r"(/2\|\.r\|\.[rq]\d\w*)$")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	97
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	98 #assert re_f.match("demo/1")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	99 assert re_f.search("demo.f")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	100 assert re_f.search("demo.s1")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	101 assert re_f.search("demo.f1k")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	102 assert re_f.search("demo.p1")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	103 assert re_f.search("demo.p1k")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	104 assert re_f.search("demo.p1lk")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	105 assert re_r.search("demo/2")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	106 assert re_r.search("demo.r")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	107 assert re_r.search("demo.q1")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	108 assert re_r.search("demo.q1lk")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	109 assert not re_r.search("demo/1")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	110 assert not re_r.search("demo.f")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	111 assert not re_r.search("demo.p")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	112 assert not re_f.search("demo/2")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	113 assert not re_f.search("demo.r")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	114 assert not re_f.search("demo.q")
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	115
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	116 count, forward, reverse, neither, pairs, singles = 0, 0, 0, 0, 0, 0
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	117 in_handle = open(input_fastq)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	118 if pairs_fastq:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	119 pairs_f_writer = fastqWriter(open(pairs_fastq, "w"), format)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	120 pairs_r_writer = pairs_f_writer
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	121 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	122 pairs_f_writer = fastqWriter(open(pairs_f_fastq, "w"), format)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	123 pairs_r_writer = fastqWriter(open(pairs_r_fastq, "w"), format)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	124 singles_writer = fastqWriter(open(singles_fastq, "w"), format)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	125 last_template, buffered_reads = None, []
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	126
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	127 for record in fastqReader(in_handle, format):
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	128 count += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	129 name = record.identifier.split(None,1)[0]
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	130 assert name[0]=="@", record.identifier #Quirk of the Galaxy parser
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	131 suffix = re_f.search(name)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	132 if suffix:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	133 #============
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	134 #Forward read
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	135 #============
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	136 template = name[:suffix.start()]
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	137 #print name, "forward", template
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	138 forward += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	139 if last_template == template:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	140 buffered_reads.append(record)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	141 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	142 #Any old buffered reads are orphans
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	143 for old in buffered_reads:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	144 singles_writer.write(old)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	145 singles += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	146 #Save this read in buffer
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	147 buffered_reads = [record]
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	148 last_template = template
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	149 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	150 suffix = re_r.search(name)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	151 if suffix:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	152 #============
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	153 #Reverse read
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	154 #============
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	155 template = name[:suffix.start()]
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	156 #print name, "reverse", template
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	157 reverse += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	158 if last_template == template and buffered_reads:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	159 #We have a pair!
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	160 #If there are multiple buffered forward reads, want to pick
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	161 #the first one (although we could try and do something more
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	162 #clever looking at the suffix to match them up...)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	163 old = buffered_reads.pop(0)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	164 pairs_f_writer.write(old)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	165 pairs_r_writer.write(record)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	166 pairs += 2
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	167 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	168 #As this is a reverse read, this and any buffered read(s) are
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	169 #all orphans
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	170 for old in buffered_reads:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	171 singles_writer.write(old)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	172 singles += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	173 buffered_reads = []
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	174 singles_writer.write(record)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	175 singles += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	176 last_template = None
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	177 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	178 #===========================
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	179 #Neither forward nor reverse
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	180 #===========================
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	181 singles_writer.write(record)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	182 singles += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	183 neither += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	184 for old in buffered_reads:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	185 singles_writer.write(old)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	186 singles += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	187 buffered_reads = []
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	188 last_template = None
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	189 if last_template:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	190 #Left over singles...
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	191 for old in buffered_reads:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	192 singles_writer.write(old)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	193 singles += 1
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	194 in_handle.close
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	195 singles_writer.close()
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	196 if pairs_fastq:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	197 pairs_f_writer.close()
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	198 assert pairs_r_writer.file.closed
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	199 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	200 pairs_f_writer.close()
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	201 pairs_r_writer.close()
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	202
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	203 if neither:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	204 print "%i reads (%i forward, %i reverse, %i neither), %i in pairs, %i as singles" \
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	205 % (count, forward, reverse, neither, pairs, singles)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	206 else:
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	207 print "%i reads (%i forward, %i reverse), %i in pairs, %i as singles" \
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	208 % (count, forward, reverse, pairs, singles)
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	209
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	210 assert count == pairs + singles == forward + reverse + neither, \
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	211 "%i vs %i+%i=%i vs %i+%i=%i" \
3a39d2053bc5 Migrated tool version 0.0.4 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	212 % (count,pairs,singles,pairs+singles,forward,reverse,neither,forward+reverse+neither)

Mercurial > repos > peterjc > fastq_paired_unpaired

annotate tools/fastq/fastq_paired_unpaired.py @ 0:3a39d2053bc5