Allow user to set max pindel threads, reinstate -force to ascat.pl

CHANGES.md

@@ -1,3 +1,8 @@
+### 1.1.4
+
+* Provide parameter to allow user to lower max pindel threads for analysis proving to have huge memory spikes
+* Add `-force` to `ascat.pl` execution, seems to have been lost along the way
+
 ### 1.1.3
 
 * Limit max threads for pindel to prevent memory explosion on noisy datasets

Dockerfile

@@ -3,7 +3,7 @@ FROM  quay.io/wtsicgp/dockstore-cgpwgs:1.1.2
 MAINTAINER  keiranmraine@gmail.com
 
 LABEL uk.ac.sanger.cgp="Cancer Genome Project, Wellcome Trust Sanger Institute" \
-      version="1.1.3" \
+      version="1.1.4" \
       description="The CGP WGS pipeline for dockstore.org"
 
 USER  root

Dockstore.cwl

@@ -22,7 +22,7 @@ dct:creator:
 
 requirements:
   - class: DockerRequirement
-    dockerPull: "quay.io/wtsicgp/dockstore-cgpwgs:1.1.3"
+    dockerPull: "quay.io/wtsicgp/dockstore-cgpwgs:1.1.4"
 
 hints:
   - class: ResourceRequirement
@@ -130,13 +130,22 @@ inputs:
       position: 11
       separate: true
 
+  pindelcpu:
+    type: int?
+    doc: "Max cpus for pindel, ignores >8"
+    default: 8
+    inputBinding:
+      prefix: -pindelcpu
+      position: 12
+      separate: true
+
   cavereads:
     type: int?
     doc: "Number of reads in a split section for CaVEMan"
     default: 350000
     inputBinding:
       prefix: -cavereads
-      position: 12
+      position: 13
       separate: true
 
 outputs:

scripts/analysisWGS.sh

@@ -71,10 +71,9 @@ if [ -z ${CPU+x} ]; then
   CPU=`grep -c ^processor /proc/cpuinfo`
 fi
 
-PINDEL_CPU=$CPU
-if [ $PINDEL_CPU -gt 8 ]; then
-  PINDEL_CPU=8
-fi
+# calculate the min of user defined, host and max we should ever allow for pindel
+min_cpu=`echo -e "$PINDEL_MAXCPU\n$CPU\n8" | sort -k1,1n | head -n 1`
+PINDEL_CPU=$min_cpu
 
 # create area which allows monitoring site to be started, not actively updated until after PRE-EXEC completes
 #cp -r /opt/wtsi-cgp/site $OUTPUT_DIR/site
@@ -252,7 +251,8 @@ do_parallel[ascat]="ascat.pl \
  -ra $ASSEMBLY \
  -pr $PROTOCOL \
  -pl ILLUMINA \
- -c $CPU"
+ -c $CPU\
+ -force"
 
 echo -e "\t[Parallel block 3] BRASS_input added..."
 do_parallel[BRASS_input]="brass.pl -j 4 -k 4 -c $CPU \

scripts/ds-wrapper.pl

@@ -12,7 +12,7 @@
 pod2usage(-verbose => 1, -exitval => 1) if(@ARGV == 0);
 
 # set defaults
-my %opts = ();
+my %opts = ('pc' => 8,);
 
 GetOptions( 'h|help' => \$opts{'h'},
             'm|man' => \$opts{'m'},
@@ -27,7 +27,8 @@
             'sp|species=s' => \$opts{'sp'},
             'as|assembly=s' => \$opts{'as'},
             'sb|skipbb' => \$opts{'sb'},
-            'cr|cavereads=i' => \$opts{'cr'}
+            'cr|cavereads=i' => \$opts{'cr'},
+            'pc|pindelcpu:i' => \$opts{'pc'}
 ) or pod2usage(2);
 
 pod2usage(-verbose => 1, -exitval => 0) if(defined $opts{'h'});
@@ -103,6 +104,7 @@
 printf $FH "BAM_MT='%s'\n", $opts{'t'};
 printf $FH "BAM_WT='%s'\n", $opts{'n'};
 printf $FH "PINDEL_EXCLUDE='%s'\n", $opts{'e'};
+printf $FH "PINDEL_MAXCPU=%d\n", $opts{'pc'};
 printf $FH "SPECIES='%s'\n", $opts{'sp'};
 printf $FH "ASSEMBLY='%s'\n", $opts{'as'};
 printf $FH "CAVESPLIT='%s'\n", $opts{'cr'};
@@ -224,6 +226,7 @@ =head1 SYNOPSIS
     -species     -sp  Species name (may require quoting)
     -assembly    -a   Reference assembly
     -skipbb      -sb  Skip Battenberg allele counts
+    -pindelcpu   -pc  Max CPUs for pindel analysis, >8 ignored [8]
 
   Other:
     -help        -h   Brief help message.
@@ -283,6 +286,11 @@ =head1 OPTION DETAILS
 
 Disables the Battenberg allele count generation
 
+=item B<-pindelcpu>
+
+Maximum parallel CPU jobs for pindel.  Useful if you have data with an extreme memory spike.
+Rarely needs setting.
+
 =back
 
 =cut