annotate repeat_annotate_custom.xml @ 4:d7bb42f254a2 draft

"planemo upload commit fa6d4eb0c804e873b60bb2ba3d0e85646c9cd764"
author petr-novak
date Mon, 14 Mar 2022 10:16:07 +0000
parents 4ea506b39297
children b53f5a456d01
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4
d7bb42f254a2 "planemo upload commit fa6d4eb0c804e873b60bb2ba3d0e85646c9cd764"
petr-novak
parents: 3
diff changeset
1 <tool id="repeat_annotate" name="Library Based Assembly Annotation" version="0.1.3" python_template_version="3.5">
0
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
2 <requirements>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
3 <requirement type="package">repeatmasker</requirement>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
4 <requirement type="package">bioconductor-rtracklayer</requirement>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
5 </requirements>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
6 <command detect_errors="exit_code"><![CDATA[
3
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
7
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
8 #if $sensitivity.value == 'default':
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
9 RepeatMasker -dir \$(pwd) '$input' -pa 32 -lib '$repeat_library' -xsmall -nolow -no_is -e ncbi
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
10 #else:
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
11 RepeatMasker -dir \$(pwd) '$input' -pa 32 -lib '$repeat_library' -xsmall -nolow -no_is -e ncbi $sensitivity
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
12 #end if
0
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
13 &&
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
14 ls -l * >&2 &&
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
15 cp `basename $input`.out $output2
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
16 &&
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
17 Rscript ${__tool_directory__}/clean_rm_output.R $output2 $output1
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
18
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
19 ]]></command>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
20 <inputs>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
21 <param type="data" name="input" format="fasta" label="Genome/ Assembly to annotate" />
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
22 <param type="data" name="repeat_library" format="fasta" label="RepeatExplorer based Library of Repetitive Sequences"
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
23 help="custom database of repetitive sequences should be provided in fasta format. Sequence header should specify repeat class:
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
24 >sequence_id#classification_level1/classification_level2/..." />
3
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
25 <param type="select" label="sensitivity" name="sensitivity" >
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
26 <option value="default" selected="true" >Default sensitivity </option>
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
27 <option value="-s">Slow search, more sensitive </option>
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
28 <option value="-q">Quick search, more sensitive </option>
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
29 </param>
0
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
30 </inputs>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
31 <outputs>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
32 <data name="output1" format="gff3" label="Repeat Annotation on ${on_string}, cleaned gff"/>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
33 <data name="output2" format="tabular" label="Raw output from RepeatMasker on ${on_string}" />
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
34 </outputs>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
35 <help><![CDATA[
3
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
36 This tools uses RepeatMasker to annotate repetitive sequences in the genome assemblies using custom library of repeats created either from RepeatExplorer output or from DANTE_LTR transposable element library.
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
37 Library of repeats can be created from RepeatExplorer output from contigs and TAREAN consensus sequences.
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
38 Fasta formatted library of repeats must contain header containing information about classification of repeats as **>sequence_id#classification_level1/classification_level2/...**
0
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
39
3
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
40 Classification in RepeatExplorer based library follows predetermined classification levels. User can however specify additional classification levels or custom classifications.
4ea506b39297 "planemo upload"
petr-novak
parents: 2
diff changeset
41 Conflicts in annotations are resolved based on classification hierarchy.
0
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
42 ]]></help>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
43 </tool>
ea6a3059a6af Uploaded
petr-novak
parents:
diff changeset
44