Merge pull request #685 from ARTbio/syncho_repenrich

back propagate repenrich2 improvement to repenrich

Author: drosofff

tools/repenrich/RepEnrich_setup.py

@@ -71,7 +71,7 @@ def import_text(filename, separator):
 genome = defaultdict(dict)
 
 for chr in g.keys():
-    genome[chr]['sequence'] = g[chr].seq
+    genome[chr]['sequence'] = str(g[chr].seq)
     genome[chr]['length'] = len(g[chr].seq)
 
 # Build a bedfile of repeatcoordinates to use by RepEnrich region_sorter
@@ -100,7 +100,7 @@ def import_text(filename, separator):
 
 # generate metagenomes and save them to FASTA files for bowtie build
 for repname in rep_coords:
-    metagenome = ''
+    genomes_list = []
     # iterating coordinate list by block of 3 (chr, start, end)
     block = 3
     for i in range(0, len(rep_coords[repname]) - block + 1, block):
@@ -109,11 +109,8 @@ def import_text(filename, separator):
         start = max(int(batch[1]) - flankingl, 0)
         end = min(int(batch[2]) + flankingl,
                   int(genome[chromosome]['length'])-1) + 1
-        metagenome = (
-            f"{metagenome}{spacer}"
-            f"{genome[chromosome]['sequence'][start:end]}"
-            )
-
+        genomes_list.append(genome[chromosome]['sequence'][start:end])
+    metagenome = spacer.join(genomes_list)
     # Create Fasta of repeat pseudogenome
     fastafilename = f"{repname}.fa"
     record = SeqRecord(Seq(metagenome), id=repname, name='', description='')

tools/repenrich/macros.xml

@@ -1,6 +1,6 @@
 <macros>
     <token name="@TOOL_VERSION@">2.31.1</token>
-    <token name="@VERSION_SUFFIX@">1</token>
+    <token name="@VERSION_SUFFIX@">2</token>
     <token name="@PROFILE@">23.0</token>
 
     <xml name="repenrich_requirements">

tools/repenrich/repenrich.xml

@@ -34,7 +34,7 @@
             #end if
         #end if
         ln -f -s '$genome' 'genome.fa' &&
-        bowtie-build '$genome' genome &&
+        bowtie-build --threads \${GALAXY_SLOTS:-1} '$genome' genome &&
         python $__tool_directory__/RepEnrich_setup.py
             --annotation_file '$repeatmasker'
             --genomefasta 'genome.fa'