Mercurial > repos > nick > sequence_content_trimmer

comparison trimmer.xml @ 1:7ef568cbf13b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 9a5d91c35d21411a0733cd233ca588ff358aabd8-dirty"
author nick
date Fri, 27 May 2022 22:44:01 +0000
parents c824894e0827
children a8fd651dcbd9
comparison
equal deleted inserted replaced
0:c824894e0827 1:7ef568cbf13b
1 <tool id="sequence_content_trimmer" version="0.1" name="Sequence Content Trimmer"> 1 <?xml version="1.0"?>
2 <tool id="sequence_content_trimmer" version="0.2.1" name="Sequence Content Trimmer">
2 <description>trim reads based on certain bases</description> 3 <description>trim reads based on certain bases</description>
3 <command interpreter="python"> 4 <command detect_errors="exit_code"><![CDATA[
4 trimmer.py $input1 5 #if $paired.is_paired and (('fasta' in $input1.extension and 'fastq' in $input2.extension) or \
5 #if $paired.is_paired: 6 ('fastq' in $input1.extension and 'fasta' in $input2.extension))
6 $input2 $output1 $output2 7 echo 'Both input files must be either fastq or fasta (no mixing the two).' >&2
7 #if ('fasta' in $input1.extension and 'fastq' in $input2.extension) or ('fastq' in $input1.extension and 'fasta' in $input2.extension) 8 #else
8 --error 'Both input files must be either fastq or fasta (no mixing the two).' 9 trimmer.py $input1
10 #if $paired.is_paired:
11 $input2 $output1 $output2
12 #end if
13 #if $input1.extension in ('fastq', 'fastqsanger', 'fastqillumina', 'fastqsolexa')
14 -f fastq
15 #elif $input1.extension == 'fasta'
16 -f fasta
17 #else
18 -f $input1.extension
19 #end if
20 -b $bases -t $thres -w $win_len $invert
21 #if $min_len.has_min_len:
22 -m $min_len.value
23 #end if
24 #if not $paired.is_paired:
25 > $output1
9 #end if 26 #end if
10 #end if 27 #end if
11 #if $input1.extension == 'fastq' or $input1.extension == 'fastqsanger' or $input1.extension == 'fastqillumina' or $input1.extension == 'fastqsolexa' 28 ]]>
12 -f fastq
13 #elif $input1.extension == 'fasta'
14 -f fasta
15 #else
16 -f $input1.extension
17 #end if
18 -b $bases -t $thres -w $win_len $invert
19 #if $min_len.has_min_len:
20 -m $min_len.value
21 #end if
22 #if not $paired.is_paired:
23 &gt; $output1
24 #end if
25 </command> 29 </command>
26 <inputs> 30 <inputs>
27 <conditional name="paired"> 31 <conditional name="paired">
28 <param name="is_paired" type="select" label="Paired reads?"> 32 <param name="is_paired" type="select" label="Paired reads?">
29 <option value="" selected="True">Unpaired</option> 33 <option value="" selected="True">Unpaired</option>
36 <when value=""> 40 <when value="">
37 <param name="input1" type="data" format="fasta,fastq" label="Input reads"/> 41 <param name="input1" type="data" format="fasta,fastq" label="Input reads"/>
38 </when> 42 </when>
39 </conditional> 43 </conditional>
40 <param name="bases" type="text" value="N" label="Bases to filter on"/> 44 <param name="bases" type="text" value="N" label="Bases to filter on"/>
41 <param name="thres" type="float" value="0.5" min="0" max="1" label="Frequency threshold" help="trim when the frequency of filter bases (or non-filter bases, if inverting) exceeds this value."/> 45 <param name="thres" type="float" value="0.5" min="0" max="1" label="Frequency threshold" help="Trim when the frequency of filter bases (or non-filter bases, if inverting) exceeds this value."/>
42 <param name="win_len" type="integer" value="10" min="1" label="Size of the window"/> 46 <param name="win_len" type="integer" value="10" min="1" label="Size of the window"/>
43 <param name="invert" type="boolean" truevalue="--invert" falsevalue="" checked="False" label="Invert filter bases" help="Trim when the frequency of bases NOT in the &quot;filter bases&quot; list exceeds the threshold."/> 47 <param name="invert" type="boolean" truevalue="--invert" falsevalue="" checked="False" label="Invert filter bases" help="Trim when the frequency of bases NOT in the &quot;filter bases&quot; list exceeds the threshold."/>
44 <conditional name="min_len"> 48 <conditional name="min_len">
45 <param name="has_min_len" type="boolean" truevalue="true" falsevalue="" checked="False" label="Set a minimum read length"/> 49 <param name="has_min_len" type="boolean" truevalue="true" falsevalue="" checked="False" label="Set a minimum read length"/>
46 <when value="true"> 50 <when value="true">
47 <param name="value" type="integer" value="10" min="0" label="Minimum read length" help="Reads trimmed to less than this length will be omitted from the output. Pairs will be preserved: both must exceed this threshold to be kept."/> 51 <param name="value" type="integer" value="10" min="0" label="Minimum read length" help="Reads trimmed to less than this length will be omitted from the output. Pairs will be preserved: both must exceed this threshold to be kept."/>
48 </when> 52 </when>
49 </conditional> 53 </conditional>
50 </inputs> 54 </inputs>
51 <outputs> 55 <outputs>
52 <data name="output1" format_source="input1"/> 56 <data name="output1" format_source="input1" label="$tool.name on $on_string"/>
53 <data name="output2" format_source="input2"> 57 <data name="output2" format_source="input2" label="$tool.name on $on_string (mate 2)">
54 <filter>paired['is_paired']</filter> 58 <filter>paired['is_paired']</filter>
55 </data> 59 </data>
56 </outputs> 60 </outputs>
57 61
58 <help> 62 <help>
64 This tool trims the 3' ends of reads based on the presence of the given bases. For instance, trim when N's are encountered or when the GC content exceeds a certain frequency. 68 This tool trims the 3' ends of reads based on the presence of the given bases. For instance, trim when N's are encountered or when the GC content exceeds a certain frequency.
65 69
66 70
67 .. class:: infomark 71 .. class:: infomark
68 72
73 **How it works**
74
75 This will slide along the read with a window, and trim once the frequency of filter bases exceeds the frequency threshold (unless "Invert filter bases" is enabled, in which case it will trim once non-filter bases exceed the threshold).
76
77 The trim point will be just before the first (leftmost) filter base in the final window (the one where the frequency exceeded the threshold).
78
79
80 .. class:: infomark
81
69 **Input** 82 **Input**
70 83
71 The inputs can be in the following formats: fasta, fastq, fastqsanger, fastqillumina, and fastqsolexa. Both must be either a fasta or fastq type (no mixing fastq and fasta). 84 The inputs can be in the following formats: fasta, fastq, fastqsanger, fastqillumina, and fastqsolexa. Both must be either a fasta or fastq type (no mixing fastq and fasta).
72 85
73 </help> 86 </help>