add review changes

galaxyproject · Dec 23, 2024 · fecf9b3 · fecf9b3
1 parent 6a9629b
commit fecf9b3
Show file tree

Hide file tree

Showing 3 changed files with 64 additions and 32 deletions.
diff --git a/tools/fairy/fairy_cov.xml b/tools/fairy/fairy_cov.xml
@@ -7,23 +7,28 @@
     <command detect_errors="exit_code">
         <![CDATA[
 
-            ln -s '$contig' '$contig.element_identifier' &&
-            ln -s '$bcsp_file' '$bcsp_file.element_identifier' &&
+            #import re
+
+            #set $file = re.sub('[^\s\w\-\\.]', '_', str($contig.element_identifier))
+            #set $bcsp = re.sub('[^\s\w\-\\.]', '_', str($bcsp_file.element_identifier))
+
+            ln -s '$contig' '$file' &&
+            ln -s '$bcsp_file' '$bcsp' &&
 
             fairy coverage
-            '$contig.element_identifier'
-            '$bcsp_file.element_identifier'
-            -t \${GALAXY_SLOTS:-8}
+            '$file'
+            '$bcsp'
+            -t "\${GALAXY_SLOTS:-3}"
             -m ${minimum_ani}
             -M ${min_number_kmers}
             -c ${c}
-            -k ${k.value}
+            -k ${k}
             --min-spacing ${min_spacing}
             ${full_contig_name}
-            #if $output_type.value == 'semi':
+            #if $output_type == 'semi':
                 --aemb-format
             #end if
-            #if $output_type.value == 'max':
+            #if $output_type == 'max':
                 --maxbin-format
             #end if
             -o '$output'
@@ -34,7 +39,7 @@
         <param name="contig" type="data" format="fasta,fasta.gz" label="Input fasta contig file" help="Input the raw fasta contig file. It can be gzip!"/>
         <param name="bcsp_file" type="data" format="bcsp" label="Input the pre-sketched file (.bcsp file)" help="This file will be generated with the fairy sketch tool."/>
         <param argument="--minimum-ani" type="integer" optional="true" min="0" max="100" value="95" label="Set minimum ANI" help="Set the minimum adjusted ANI for the coverage calculation. CARE: only adjust it when you know what it does!"/>
-        <param argument="--min-number-kmers" type="integer" value="8" optional="true" label="Genome filter" help="Set the number for exclude genomes with less than this number of sampled k-mers."/>
+        <param argument="--min-number-kmers" type="integer" value="8" optional="true" label="Genome filter" help="Set a filter to filter out genomes with less then x k-mer sampled."/>
         <param argument="-c" type="integer" value="50" optional="true" label="Set subsampling rate" help="This value does not interact with the .bcsp file which was used as input."/>
         <param argument="-k" type="select" label="Select k-mer size" help="This value does not interact with the .bcsp file which was used as input.">
             <option value="31">31</option>

diff --git a/tools/fairy/fairy_sketch.xml b/tools/fairy/fairy_sketch.xml
@@ -6,32 +6,43 @@
     <expand macro="requirements"/>
     <command detect_errors="exit_code">
         <![CDATA[
+        
+            #import re
 
-        mkdir -p res &&
+            mkdir -p res &&
 
-        #if $input.is_select == "single":
-            #set $filename = $reads.element_identifier + '.bcsp'
-            ln -s '$reads' '$reads.element_identifier' &&
-        #else
-            #set $filename = $first_pairs.element_identifier + '.paired.bcsp'
-            ln -s '$first_pairs' '$first_pairs.element_identifier' &&
-            ln -s '$second_pairs' '$second_pairs.element_identifier' &&
-        #end if
+            #if $input.is_select == "single":
+                #set $file = re.sub('[^\s\w\-\\.]', '_', str($reads.element_identifier))
+                #set $filename = $file + '.bcsp'
+                ln -s '$reads' '$file' &&
+            #else if $input.is_select == 'pair':
+                #set $file_1 = re.sub('[^\s\w\-\\.]', '_', str($first_pairs.element_identifier))
+                #set $file_2 = re.sub('[^\s\w\-\\.]', '_', str($second_pairs.element_identifier))
+                #set $filename = $file_1 + '.paired.bcsp'
+                ln -s '$first_pairs' '$file_1' &&
+                ln -s '$second_pairs' '$file_2' &&
+            #else
+                #set $file_1 = re.sub('[^\s\w\-\\.]', '_', str($paired_collection.forward.element_identifier))
+                #set $file_2 = re.sub('[^\s\w\-\\.]', '_', str($paired_collection.reverse.element_identifier))
+                #set $filename = $file_1 + '.paired.bcsp'
+                ln -s '$paired_collection.forward' '$file_1' &&
+                ln -s '$paired_collection.reverse' '$file_2' &&
+            #end if
 
-        fairy sketch
-        -t \${GALAXY_SLOTS:-8}
-        -c ${c}
-        -k ${k.value}
-        -d 'res'
-        #if $input.is_select == "single":
-            -r '$reads.element_identifier'
-        #else
-            -1 '$first_pairs.element_identifier'
-            -2 '$second_pairs.element_identifier'
-        #end if
-        &&
+            fairy sketch
+            -t "\${GALAXY_SLOTS:-3}"
+            -c ${c}
+            -k ${k}
+            -d 'res'
+            #if $input.is_select == "single":
+                -r '$file'
+            #else
+                -1 '$file_1'
+                -2 '$file_2'
+            #end if
+            &&
 
-        cp './res/${filename}' $output
+            cp './res/${filename}' $output
 
         ]]>
     </command>
@@ -40,14 +51,18 @@
             <param name="is_select" type="select" label="Single or paired-end reads">
                 <option value="single">Single</option>
                 <option value="pair">Paired</option>
+                <option value="paired_collection">Paired collection</option>
             </param>
             <when value="single">
                 <param argument="--reads" type="data" format="fastqsanger,fasta,fastq,fasta.gz,fastq.gz" label="Input single-end reads"/>
             </when>
             <when value="pair">
                 <param argument="--first_pairs" type="data" format="fastqsanger,fasta,fastq,fasta.gz,fastq.gz" label="Input first paired-end reads"/>
                 <param argument="--second_pairs" type="data" format="fastqsanger,fasta,fastq,fasta.gz,fastq.gz" label="Input second paired-end reads"/>
-            </when>  
+            </when>
+            <when value="paired_collection">
+                <param name="paired_collection" format="fastqsanger,fasta,fastq,fasta.gz,fastq.gz" type="data_collection" collection_type="paired" label="Input paired collection reads"/>
+            </when>
         </conditional>
         <param argument="-c" type="integer" value="50" optional="true" label="Set the subsampling rate"/>
         <param argument="-k" type="select" label="Select k-mer size">
@@ -74,6 +89,18 @@
             </conditional>
             <output name="output" file="test_paired_1.fq.gz.paired.bcsp"/>
         </test>
+        <test>
+            <conditional name="input">
+                <param name="is_select" value="paired_collection"/>
+                <param name="paired_collection">
+                    <collection type="paired">
+                        <element name="forward" value="test_paired_1.fq.gz" ftype="fastq.gz" />
+                        <element name="reverse" value="test_paired_2.fq.gz" ftype="fastq.gz" />
+                    </collection>
+                </param>    
+            </conditional>
+            <output name="output" file="forward.paired.bcsp"/>
+        </test>
     </tests>
     <help>
         <![CDATA[

diff --git a/tools/fairy/test-data/forward.paired.bcsp b/tools/fairy/test-data/forward.paired.bcsp