Mercurial > repos > bcclaywell > microbiome_pplacer_suite
changeset 3:221a97bf686d draft
Uploaded
author | bcclaywell |
---|---|
date | Wed, 25 Mar 2015 16:39:10 -0400 |
parents | ce6db18f5fd3 |
children | 5f0648f799a9 |
files | alignment-wrapper.sh alignment.xml preprocessing-wrapper.sh preprocessing.xml |
diffstat | 4 files changed, 64 insertions(+), 22 deletions(-) [+] |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/alignment-wrapper.sh Wed Mar 25 16:39:10 2015 -0400 @@ -0,0 +1,14 @@ +#!/bin/bash + +source $1 + +# adapted from yapp/bin/refpkg_align +ref_sto=$(taxit rp ${REFPKG} aln_sto) +profile=$(taxit rp ${REFPKG} profile) + +sto=$(mktemp -u).sto + +cmalign --cpu ${GALAXY_SLOTS:-4} -o "$sto" --sfile "${ALIGNED_SCORES}" --noprob --dnaout "$profile" "${INPUT_SEQS}" | grep -E '^#' + +esl-alimerge --dna --outformat afa "$ref_sto" "$sto" | \ + seqmagick convert --output-format fasta --dash-gap - "${ALIGNED_SEQS}"
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/alignment.xml Wed Mar 25 16:39:10 2015 -0400 @@ -0,0 +1,47 @@ +<tool id="PHYLO_alignment" name="Align query sequences" version="1.0.0"> + <description>with sequences in a reference package</description> + <requirements> + <requirement type="package">yapp_env</requirement> + </requirements> + <macros> + <import>macros.xml</import> + </macros> + <version_command>echo "guppy $(guppy --version)"</version_command> + <command interpreter="bash"> + alignment-wrapper.sh ${config} + </command> + <stdio> + <expand macro="basic_errors"/> + </stdio> + <inputs> + <param name="refpkg" type="data" format="refpkg" label="Reference package"/> + <param name="input_seqs" type="data" format="fasta" label="Input sequences"/> + </inputs> + <outputs> + <data name="aligned_seqs" format="fasta" label="Aligned sequences"/> + <data name="aligned_scores" format="txt" label="Alignment scores"/> + </outputs> + <configfiles> + <configfile name="config"> +REFPKG="${refpkg.extra_files_path}" +INPUT_SEQS="${input_seqs}" + +ALIGNED_SEQS="${aligned_seqs}" +ALIGNED_SCORES="${aligned_scores}" + </configfile> + </configfiles> + <!-- The contents of the help tag is parsed as reStructuredText. Please see + help-template.rst for examples of commonly-used sections in other Galaxy + tools. --> + <help> + +.. class:: infomark + +**What it does** + +This tool aligns query sequences with the reference sequences used to make the +reference tree contained in the reference package and then merges the query and +reference sequences. + + </help> +</tool>
--- a/preprocessing-wrapper.sh Thu Feb 26 19:31:20 2015 -0500 +++ b/preprocessing-wrapper.sh Wed Mar 25 16:39:10 2015 -0400 @@ -7,14 +7,3 @@ --deduplicated-sequences-file ${DEDUP_INFO} \ ${INPUT_SEQS} \ ${DEDUP_SEQS} - -# adapted from yapp/bin/refpkg_align -ref_sto=$(taxit rp ${REFPKG} aln_sto) -profile=$(taxit rp ${REFPKG} profile) - -sto=$(mktemp -u).sto - -cmalign --cpu ${GALAXY_SLOTS:-4} -o "$sto" --sfile "${ALIGNED_SCORES}" --noprob --dnaout "$profile" "${DEDUP_SEQS}" | grep -E '^#' - -esl-alimerge --dna --outformat afa "$ref_sto" "$sto" | \ - seqmagick convert --output-format fasta --dash-gap - "${ALIGNED_SEQS}"
--- a/preprocessing.xml Thu Feb 26 19:31:20 2015 -0500 +++ b/preprocessing.xml Wed Mar 25 16:39:10 2015 -0400 @@ -1,5 +1,5 @@ -<tool id="PHYLO_preprocessing" name="Preprocess sequences" version="2.0.0"> - <description>in preparation for phylogenetic placement</description> +<tool id="PHYLO_preprocessing" name="Deduplicate sequences" version="3.0.0"> + <description>in preparation for alignment and placement</description> <requirements> <requirement type="package">yapp_env</requirement> </requirements> @@ -14,26 +14,20 @@ <expand macro="basic_errors"/> </stdio> <inputs> - <param name="refpkg" type="data" format="refpkg" label="Reference package"/> <param name="input_seqs" type="data" format="fasta" label="Input sequences"/> <param name="split_map" type="data" format="csv" label="Read-to-specimen map"/> </inputs> <outputs> <data name="dedup_seqs" format="fasta" label="Deduplicated sequences"/> <data name="dedup_info" format="csv" label="Deduplication info"/> - <data name="aligned_seqs" format="fasta" label="Aligned sequences"/> - <data name="aligned_scores" format="txt" label="Alignment scores"/> </outputs> <configfiles> <configfile name="config"> -REFPKG="${refpkg.extra_files_path}" INPUT_SEQS="${input_seqs}" SPLIT_MAP="${split_map}" DEDUP_SEQS="${dedup_seqs}" DEDUP_INFO="${dedup_info}" -ALIGNED_SEQS="${aligned_seqs}" -ALIGNED_SCORES="${aligned_scores}" </configfile> </configfiles> <!-- The contents of the help tag is parsed as reStructuredText. Please see @@ -45,9 +39,7 @@ **What it does** -This tool aligns query sequences with the reference sequences used to make the -reference tree contained in the reference package and then merges the query and -reference sequences. +This tool deduplicates input sequences in preparation for alignment and placement. </help> </tool>