diff repeatmasker.xml @ 4:04f5c3d7448e draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/repeat_masker commit cbba6947f1380751e1db3fa5b043af630523fd86

--- a/repeatmasker.xml	Wed May 02 20:18:11 2018 -0400
+++ b/repeatmasker.xml	Fri May 04 07:59:50 2018 -0400
@@ -1,4 +1,4 @@
-<tool id="repeatmasker_wrapper" name="RepeatMasker" version="4.0.7" profile="17.01">
+<tool id="repeatmasker_wrapper" name="RepeatMasker" version="4.0.7+galaxy1" profile="17.01">
   <description>RepeatMasker</description>
 
   <requirements>
@@ -11,7 +11,7 @@
     export REPEATMASKER_LIB_DIR=\$(pwd)/lib &&
       for file in \$(ls \$RM_LIB_PATH) ; do  ln -s \$RM_LIB_PATH/\$file lib/\$file ; done &&
     #if $repeat_source.source_type == "repbase":
-      cp '${repeat_source.repbase_file}' lib/RMRBSeqs.embl &&
+      cp '${repeat_source.repbase_file}' 'lib/${repeat_source.repbase_file_name}' &&
     #end if
     ln -s '${input_fasta}' rm_input.fasta &&
     RepeatMasker -dir \$(pwd)
@@ -26,31 +26,31 @@
       #end if
     #end if
     -parallel \${GALAXY_SLOTS:-1}
-    '${gff}'
-    '${ignore_n_stretches}'
-    '${advanced.is_only}'
-    '${advanced.is_clip}'
-    '${advanced.no_is}'
-    '${advanced.rodspec}'
-    '${advanced.primspec}'
-    '${advanced.nolow}'
-    '${advanced.noint}'
-    '${advanced.norna}'
-    '${advanced.alu}'
-    '${advanced.div}'
-    '${advanced.search_speed}'
-    '${advanced.frag}'
-    '${advanced.maxsize}'
-    #if $advanced.gc is not None:
-      '${advanced.gc}'
+    ${gff}
+    ${excln}
+    ${advanced.is_only}
+    ${advanced.is_clip}
+    ${advanced.no_is}
+    ${advanced.rodspec}
+    ${advanced.primspec}
+    ${advanced.nolow}
+    ${advanced.noint}
+    ${advanced.norna}
+    ${advanced.alu}
+    ${advanced.div}
+    ${advanced.search_speed}
+    -frag ${advanced.frag}
+    ## -maxsize ${advanced.maxsize}
+    #if str($advanced.gc):
+      -gc ${advanced.gc}
     #end if
-    '${advanced.gccalc}'
-    '${advanced.nocut}'
-    '${advanced.keep_alignments}'
-    '${advanced.invert_alignments}'
-    '${advanced.xout}'
-    '${advanced.xsmall}'
-    '${advanced.poly}'
+    ${advanced.gccalc}
+    ${advanced.nocut}
+    ${advanced.keep_alignments}
+    ${advanced.invert_alignments}
+    ${advanced.xout}
+    ${advanced.xsmall}
+    ${advanced.poly}
     rm_input.fasta &&
     #if $advanced.is_only != '-is_only':
       mv rm_input.fasta.masked '${output_masked_genome}' &&
@@ -79,6 +79,7 @@
       </param>
       <when value="repbase">
         <param name="repbase_file" type="data" format="embl" label="RepBase (RMRBSeqs.embl) file" />
+        <param name="repbase_file_name" type="hidden" value="RMRBSeqs.embl"/> <!-- This is an ugly hack to allow testing with a fake repbase -->
         <conditional name="species_source">
           <param label="Select species name from a list?" name="species_from_list" type="select">
             <option value="yes" selected="true">Yes</option>
@@ -126,7 +127,7 @@
       </when>
     </conditional>
     <param type="boolean" argument="-gff" truevalue="-gff" falsevalue="" label="Output annotation of repeats in GFF format" checked="false" />
-    <param name="ignore_n_stretches" type="boolean" argument="-excln" falsevalue="" label="Ignore stretches of Ns when computing statistics" checked="true" help="Scaffolds are sometimes joined with stretches of 25 or more Ns. This option ignores them when calculating repeat statistics" />
+    <param argument="-excln" type="boolean" truevalue="-excln" falsevalue="" label="Ignore stretches of Ns when computing statistics" checked="true" help="Scaffolds are sometimes joined with stretches of 25 or more Ns. This option ignores them when calculating repeat statistics" />
     <section name="advanced" title="Advanced options" expanded="false">
       <param argument="-is_only" type="boolean" truevalue="-is_only" falsevalue="" checked="false" label="Only clip E coli insertion elements" />
       <param argument="-is_clip" type="boolean" truevalue="-is_clip" falsevalue="" checked="false" label="Clip IS elements before analysis" help="Normally RepeatMasker will report on IS element, with this option selected it will clip them before analysis" />
@@ -145,8 +146,9 @@
         <option value="-s">Slow (0-5% more sensitive, 2.5 times slowdown)</option>
       </param>
       <param type="integer" argument="-frag" value="40000" label="Maximum contiguous sequence searched" help="Maximum length of sequencing that is search without fragmenting" />
-      <param type="integer" argument="-maxsize" value="4000000" label="Maximum length for IS or repeat clipped sequences" />
-      <param type="integer" argument="-gc" optional="True" label="Select matrices for this GC%" help="Valid values are a percentage or -1 to choose the default" />
+      <!-- -maxsize option is in the help, but not in the code of repeatmasker-->
+      <!--param type="integer" argument="-maxsize" value="4000000" label="Maximum length for IS or repeat clipped sequences" /-->
+      <param type="integer" argument="-gc" optional="true" label="Select matrices for this GC%" help="Valid values are a percentage or -1 to choose the default" />
       <param type="boolean" argument="-gccalc" truevalue="-gcccalc" falsevalue="" checked="false" label="Calculate GC % for all sequences" help="By default RepeatMasker skips calculating GC % for small sequences" />
       <param type="boolean" argument="-nocut" truevalue="-nocut" falsevalue="" checked="false" label="Skips cutting of repeats" />
       <param name="xout" type="boolean" argument="-x" truevalue="-x" falsevalue="" checked="false" label="Mask with X instead of N characters" />
@@ -203,6 +205,17 @@
       <output name="output_polymorphic" file="small.fasta.poly" />
       <output name="output_gff" file="small.fasta.gff" lines_diff="4" />
     </test>
+    <test expect_num_outputs="4">
+      <param name="input_fasta" value="small.fasta" ftype="fasta" />
+      <param name="source_type" value="repbase" />
+      <param name="repbase_file" value="fake_repbase.embl" />
+      <param name="repbase_file_name" value="fake.embl" />
+      <param name="species_list" value="anopheles" />
+      <output name="output_masked_genome" file="small.fasta.masked" />
+      <output name="output_table" file="small_repbase.fasta.stats" lines_diff="2" />
+      <output name="output_repeat_catalog" file="small.fasta.cat" />
+      <output name="output_log" file="small_repbase.fasta.log" />
+    </test>
   </tests>
   <help><![CDATA[
 RepeatMasker is a program that screens DNA for interspersed repeats and low