Mercurial > repos > jjohnson > samtools_split_by_chrom

annotate samtools_split_by_chrom.xml @ 1:6fb39843d37d draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
author jjohnson
date Wed, 22 Mar 2017 17:22:35 -0400
parents a30dd3c77b30
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
1 <tool id="samtools_split_by_chrom" name="Split BAM by Chromosome" version="2.0">
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
2 <description>into collection</description>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
3 <macros>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
4 <import>macros.xml</import>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
5 </macros>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
6 <!-- <code file="samtools_slice_options.py"/> -->
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
7 <expand macro="requirements"></expand>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
8 <expand macro="stdio"></expand>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
9 <expand macro="version_command"></expand>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
10 <command>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
11 <![CDATA[
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
12 #import re
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
13 #set $name = $re.sub('\.bam$','',$input_bam.name)
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
14 #if str($refs) != 'None':
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
15 #set ref_list = str($refs).split(",")
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
16 #else
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
17 #set ref_list = [$re.sub('^.*__sq__(.+)__sq__.*$','\\1',n) if n.find('__sq__') >= 0 else n for n in str($input_bam.metadata.reference_names).split(',')]
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
18 #end if
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
19 mkdir -p outputs &&
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
20 ln -s "${input_bam}" temp_input.bam &&
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
21 ln -s "${input_bam.metadata.bam_index}" temp_input.bam.bai
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
22 #for $i,$ref in enumerate($ref_list):
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
23 #set $idx = "%04d" % $i
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
24 && samtools view -@ \${GALAXY_SLOTS:-1} -bh temp_input.bam ${ref} |
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
25 samtools sort -O bam -T sorted -@ \${GALAXY_SLOTS:-1} -o "outputs/${idx}-${name}.${ref}.bam" -
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
26 #end for
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
27 && ls -l outputs | awk '/bam/{fname = substr(\$NF,6); printf("%s\t%d\n", fname, \$5)}' > "ls_split_files"
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
28 ]]>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
29 </command>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
30 <inputs>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
31 <param name="input_bam" format="bam" label="Select BAM dataset to slice" type="data" />
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
32 <param name="refs" type="select" optional="True" multiple="True" label="Select references (chromosomes and contigs) you would like to restrict bam to" help="Click and type in the box above to see options. You can select multiple entries. If &quot;No options available&quot; is displayed, you need to re-detect metadata on the input dataset. See help section below.">
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
33 <options>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
34 <filter type="data_meta" ref="input_bam" key="reference_names" />
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
35 </options>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
36 </param>
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
37 <param name="show_listing" type="boolean" truevalue="yes" falsevalue="no" checked="false" label="show listing"/>
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
38 </inputs>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
39 <outputs>
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
40 <data format="tabular" name="listing" from_work_dir="ls_split_files">
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
41 <filter>show_listing</filter>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
42 <actions>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
43 <action name="column_names" type="metadata" default="name,size" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
44 </actions>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
45 </data>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
46 <collection name="output_collection" type="list" label="${input_bam.name} by chrom">
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
47 <discover_datasets pattern="\d+-(?P&lt;designation&gt;.*\.bam)" directory="outputs" format="bam" ext="bam" visible="false"/>
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
48 </collection>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
49 </outputs>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
50 <tests>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
51 <test>
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
52 <param ftype="bam" name="input_bam" value="test.bam" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
53 <param name="show_listing" value="True" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
54 <output file="listing">
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
55 <assert_contents>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
56 <has_text expression="chr1" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
57 <has_text expression="chr2" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
58 </assert_contents>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
59 </output>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
60 </test>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
61 <test>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
62 <param ftype="bam" name="input_bam" value="test.bam" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
63 <param name="refs" value="chr1,chr3" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
64 <param name="show_listing" value="True" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
65 <output file="listing">
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
66 <assert_contents>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
67 <has_text expression="chr1" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
68 <not_has_text expression="chr2" />
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
69 </assert_contents>
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
70 </output>
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
71 </test>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
72 </tests>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
73 <help>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
74 <![CDATA[
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
75
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
76 **What it does**
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
77
1
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
78 Splits an input BAM dataset to a dataset collection of individual chromosome bam files.
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
79
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
80 This dataset collection can be passed to a galaxy tool that takes a single bam input
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
81 in order to split the bam processing across multiple jobs.
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
82
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
83 A suggested use case:
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
84
6fb39843d37d planemo upload commit f3f0bef4a450aafab3c6b05a27647471f93b22f3
jjohnson
parents: 0
diff changeset
85 hisat -> samtools_split_by_chrom => bcftools_mpileup => bcftools_call => bcftools_merge -> snpEff
0
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
86
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
87 This tool is based on ``samtools view`` command.
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
88
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
89 ]]>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
90 </help>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
91 <expand macro="citations"></expand>
a30dd3c77b30 planemo upload commit 7491208ca0c917a053798a48c3e54c3e30e95d92
jjohnson
parents:
diff changeset
92 </tool>