Merge pull request galaxyproject#5492 from mvdbeek/trinity_scratch_dir

Simplify and fix trinity scratch dir handling
pavanvidem · Sep 20, 2023 · ec26c47 · ec26c47
2 parents c083399 + b36dbe6
commit ec26c47
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 34 deletions.
diff --git a/tools/trinity/abundance_estimates_to_matrix.xml b/tools/trinity/abundance_estimates_to_matrix.xml
@@ -89,7 +89,7 @@
         </data>
     </outputs>
     <tests>
-        <test>
+        <test expect_num_outputs="3">
             <param name="samples" ftype="tabular" value="count/rsem_bowtie/sample_A,count/rsem_bowtie/sample_B"/>
             <param name="est_method" value="RSEM"/>
             <param name="cross_sample_norm" value="TMM"/>
@@ -112,7 +112,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="3">
             <param name="samples" ftype="tabular" value="count/rsem_bowtie2/sample_A,count/rsem_bowtie2/sample_B"/>
             <param name="est_method" value="RSEM"/>
             <param name="cross_sample_norm" value="TMM"/>
@@ -135,7 +135,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="3">
             <param name="samples" ftype="tabular" value="count/express_bowtie/sample_A,count/express_bowtie/sample_B"/>
             <param name="est_method" value="eXpress"/>
             <param name="cross_sample_norm" value="TMM"/>
@@ -158,7 +158,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="3">
             <param name="samples" ftype="tabular" value="count/rsem_bowtie/sample_A,count/rsem_bowtie/sample_B"/>
             <param name="est_method" value="RSEM"/>
             <param name="cross_sample_norm" value="UpperQuartile"/>
@@ -180,7 +180,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="2">
             <param name="samples" ftype="tabular" value="count/rsem_bowtie/sample_A,count/rsem_bowtie/sample_B"/>
             <param name="est_method" value="RSEM"/>
             <param name="cross_sample_norm" value="none"/>
@@ -197,7 +197,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="2">
             <param name="samples" ftype="tabular" value="count/salmon/sample_A,count/salmon/sample_B"/>
             <param name="est_method" value="salmon"/>
             <param name="cross_sample_norm" value="none"/>
@@ -214,7 +214,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="3">
             <param name="samples" ftype="tabular" value="count/kallisto/abundance.tsv.genes,count/kallisto/abundance_B.tsv.genes"/>
             <param name="est_method" value="kallisto"/>
             <param name="cross_sample_norm" value="TMM"/>

diff --git a/tools/trinity/align_and_estimate_abundance.xml b/tools/trinity/align_and_estimate_abundance.xml
@@ -180,7 +180,7 @@
         </data>
     </outputs>
     <tests>
-        <test>
+        <test expect_num_outputs="2">
             <param name="paired_or_single" value="paired"/>
             <param name="left_input" value="reads.left.fq.gz"/>
             <param name="right_input" value="reads.right.fq.gz"/>
@@ -203,7 +203,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="2">
             <param name="paired_or_single" value="paired"/>
             <param name="left_input" value="reads.left.fq"/>
             <param name="right_input" value="reads.right.fq"/>
@@ -226,7 +226,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="2">
             <param name="paired_or_single" value="paired"/>
             <param name="left_input" value="reads.left.fq"/>
             <param name="right_input" value="reads.right.fq"/>
@@ -249,7 +249,7 @@
                 </assert_contents>
             </output>
         </test>
-        <test>
+        <test expect_num_outputs="2">
             <param name="paired_or_single" value="paired"/>
             <param name="left_input" value="reads.left.fq"/>
             <param name="right_input" value="reads.right.fq"/>

diff --git a/tools/trinity/analyze_diff_expr.xml b/tools/trinity/analyze_diff_expr.xml
@@ -96,7 +96,7 @@
     </outputs>
     <tests>
         <!-- Test without GO enrichment analysis -->
-        <test>
+        <test expect_num_outputs="9">
             <param name="matrix" value="count/qcheck/matrix.counts.matrix"/>
             <param name="samples" value="count/samples.txt"/>
             <param name="DE_results">
@@ -125,7 +125,7 @@
             <output name="rdata" compare="sim_size" file="count/analyze_diff_expr/results.matrix.RData"/>
         </test>
         <!-- Test without GO enrichment analysis, test name of DE_results input files different from those generated by tool run_de_analysis.pl -->
-        <test>
+        <test expect_num_outputs="9">
             <param name="matrix" value="count/qcheck/matrix.counts.matrix"/>
             <param name="samples" value="count/samples.txt"/>
             <param name="DE_results">
@@ -154,7 +154,7 @@
             <output name="rdata" compare="sim_size" file="count/analyze_diff_expr/results.matrix.RData"/>
         </test>
         <!-- Test with GO enrichment analysis -->
-        <test>
+        <test expect_num_outputs="10">
             <param name="matrix" value="count/qcheck/matrix.counts.matrix"/>
             <param name="samples" value="count/samples.txt"/>
             <param name="DE_results">

diff --git a/tools/trinity/trinity.xml b/tools/trinity/trinity.xml
@@ -1,4 +1,4 @@
-<tool id="trinity" name="Trinity" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="21.05">
+<tool id="trinity" name="Trinity" version="@TOOL_VERSION@+galaxy1" profile="21.05">
     <description>de novo assembly of RNA-Seq data</description>
     <macros>
         <import>macros.xml</import>
@@ -14,12 +14,11 @@
             GALAXY_MEMORY_GB=\$((GALAXY_MEMORY_MB / 1024)) ;
         fi ;
 
-        if [ ! -z "\$TRINITY_SCRATCH_DIR" ] ; then
-            workdir=`pwd` ;
-            scratchfolder=\$(mktemp -d -p "\$TRINITY_SCRATCH_DIR");
-            emptyfolder=\$(mktemp -d -p "\$TRINITY_SCRATCH_DIR");
-            cd "\$scratchfolder" ;
-        fi ;
+        TRINITY_SCRATCH_DIR=\${TRINITY_SCRATCH_DIR:-\${TMPDIR:-\$_GALAXY_JOB_TMP_DIR}}
+        workdir=\$(pwd)
+        scratchfolder=\$(mktemp -d -p "\$TRINITY_SCRATCH_DIR");
+        emptyfolder=\$(mktemp -d -p "\$TRINITY_SCRATCH_DIR");
+        cd "\$scratchfolder" ;
 
         #if $additional_params.guided.is_guided == "yes":
             ln -s '${$additional_params.guided.genome_guided_bam}' 'localbam.bam' &&
@@ -148,22 +147,16 @@
         ## CPU and butterfly options.
         --CPU \${GALAXY_SLOTS:-4} --max_memory \${GALAXY_MEMORY_GB:-1}G --bflyHeapSpaceMax \${GALAXY_MEMORY_GB:-1}G --bfly_opts '-V 10 --stderr'
 
-        ## > $trinity_log 2>&1
-
         &&
 
         ## Trinity can create millions of files in the same directory, so the cleaning task makes use of rsync
         ## for ensuring better performances.
         ## see: https://web.archive.org/web/20130929001850/http://linuxnote.net/jianingy/en/linux/a-fast-way-to-remove-huge-number-of-files.html
-        if [ ! -z "\$TRINITY_SCRATCH_DIR" ] ; then
-            mkdir -p "\$workdir/trinity_out_dir";
-            cp -p trinity_out_dir/Trinity* "\$workdir/trinity_out_dir";
-            cd "\$TRINITY_SCRATCH_DIR";
-            rsync -a --delete "\$emptyfolder/" "\$scratchfolder/";
-            rmdir "\$emptyfolder" "\$scratchfolder/";
-            cd "\$workdir";
-        fi ;
-
+        rsync -a --delete "\$emptyfolder/" "\$scratchfolder/" --exclude=trinity_out_dir.Trinity.fasta --exclude=trinity_out_dir.Trinity.fasta.gene_trans_map;
+        mv "\$scratchfolder/trinity_out_dir.Trinity.fasta" '$assembled_transcripts';
+        mv "\$scratchfolder/trinity_out_dir.Trinity.fasta.gene_trans_map" '$gene_to_trans';
+        cd "\$workdir";
+        rmdir "\$emptyfolder" "\$scratchfolder"
     ]]></command>
     <inputs>
         <conditional name="pool">
@@ -235,8 +228,8 @@
         </section>
     </inputs>
     <outputs>
-        <data name="assembled_transcripts" format="fasta" label="${tool.name} on ${on_string}: Assembled Transcripts" from_work_dir="trinity_out_dir.Trinity.fasta"/>
-        <data name="gene_to_trans" format="tabular" label="${tool.name} on ${on_string}: Gene to transcripts map" from_work_dir="trinity_out_dir.Trinity.fasta.gene_trans_map"/>
+        <data name="assembled_transcripts" format="fasta" label="${tool.name} on ${on_string}: Assembled Transcripts"/>
+        <data name="gene_to_trans" format="tabular" label="${tool.name} on ${on_string}: Gene to transcripts map"/>
     </outputs>
     <tests>
         <test>