|
3
|
1 <tool id="repeat_annotate" name="Library Based Assembly Annotation" version="0.1.2" python_template_version="3.5">
|
|
0
|
2 <requirements>
|
|
|
3 <requirement type="package">repeatmasker</requirement>
|
|
|
4 <requirement type="package">bioconductor-rtracklayer</requirement>
|
|
|
5 </requirements>
|
|
|
6 <command detect_errors="exit_code"><![CDATA[
|
|
3
|
7
|
|
|
8 #if $sensitivity.value == 'default':
|
|
|
9 RepeatMasker -dir \$(pwd) '$input' -pa 32 -lib '$repeat_library' -xsmall -nolow -no_is -e ncbi
|
|
|
10 #else:
|
|
|
11 RepeatMasker -dir \$(pwd) '$input' -pa 32 -lib '$repeat_library' -xsmall -nolow -no_is -e ncbi $sensitivity
|
|
|
12 #end if
|
|
0
|
13 &&
|
|
|
14 ls -l * >&2 &&
|
|
|
15 cp `basename $input`.out $output2
|
|
|
16 &&
|
|
|
17 Rscript ${__tool_directory__}/clean_rm_output.R $output2 $output1
|
|
|
18
|
|
|
19 ]]></command>
|
|
|
20 <inputs>
|
|
|
21 <param type="data" name="input" format="fasta" label="Genome/ Assembly to annotate" />
|
|
|
22 <param type="data" name="repeat_library" format="fasta" label="RepeatExplorer based Library of Repetitive Sequences"
|
|
|
23 help="custom database of repetitive sequences should be provided in fasta format. Sequence header should specify repeat class:
|
|
|
24 >sequence_id#classification_level1/classification_level2/..." />
|
|
3
|
25 <param type="select" label="sensitivity" name="sensitivity" >
|
|
|
26 <option value="default" selected="true" >Default sensitivity </option>
|
|
|
27 <option value="-s">Slow search, more sensitive </option>
|
|
|
28 <option value="-q">Quick search, more sensitive </option>
|
|
|
29 </param>
|
|
0
|
30 </inputs>
|
|
|
31 <outputs>
|
|
|
32 <data name="output1" format="gff3" label="Repeat Annotation on ${on_string}, cleaned gff"/>
|
|
|
33 <data name="output2" format="tabular" label="Raw output from RepeatMasker on ${on_string}" />
|
|
|
34 </outputs>
|
|
|
35 <help><![CDATA[
|
|
3
|
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.
|
|
|
37 Library of repeats can be created from RepeatExplorer output from contigs and TAREAN consensus sequences.
|
|
|
38 Fasta formatted library of repeats must contain header containing information about classification of repeats as **>sequence_id#classification_level1/classification_level2/...**
|
|
0
|
39
|
|
3
|
40 Classification in RepeatExplorer based library follows predetermined classification levels. User can however specify additional classification levels or custom classifications.
|
|
|
41 Conflicts in annotations are resolved based on classification hierarchy.
|
|
0
|
42 ]]></help>
|
|
|
43 </tool>
|
|
|
44
|
