Merge pull request #421 from jlac/activeDev

reduced verbosity of VarDict and added MutSigCV for additional callers

Rules/mutsig_merged.rl

@@ -0,0 +1,30 @@
+num = sum(1 for line in open('pairs'))
+
+if num > 10:
+  if config['project']['annotation'] == "hg19":
+    rule mutsig_merged:
+        input: config['project']['workpath']+"/merged_somatic_variants/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/merged_somatic_variants/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_merged"
+        shell: "module load MutSig; MutSigCV {input} $MUTSIG_REF/exome_full192.coverage.txt $MUTSIG_REF/gene.covariates.txt merged_somatic_variants/mutsigCV_out/somatic $MUTSIG_REF/mutation_type_dictionary_file.txt $MUTSIG_REF/chr_files_hg19"
+
+  if config['project']['annotation'] == "hg38":
+    rule mutsig_merged:
+        input: config['project']['workpath']+"/merged_somatic_variants/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/merged_somatic_variants/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_merged"
+        shell: "echo \"null\" > {output.genes}"
+
+  elif config['project']['annotation'] == "mm10":
+    rule mutsig_merged:
+        input: config['project']['workpath']+"/merged_somatic_variants/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/merged_somatic_variants/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_merged"
+        shell: "echo \"null\" > {output.genes}"
+
+else:
+    rule mutsig_merged:
+        input: config['project']['workpath']+"/merged_somatic_variants/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/merged_somatic_variants/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_merged"
+        shell: "echo \"null\" > {output.genes}"

Rules/mutsig_vardict.rl

@@ -0,0 +1,30 @@
+num = sum(1 for line in open('pairs'))
+
+if num > 10:
+  if config['project']['annotation'] == "hg19":
+    rule mutsig_vardict:
+        input: config['project']['workpath']+"/vardict_out/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/vardict_out/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_vardict"
+        shell: "module load MutSig; MutSigCV {input} $MUTSIG_REF/exome_full192.coverage.txt $MUTSIG_REF/gene.covariates.txt vardict_out/mutsigCV_out/somatic $MUTSIG_REF/mutation_type_dictionary_file.txt $MUTSIG_REF/chr_files_hg19"
+
+  if config['project']['annotation'] == "hg38":
+    rule mutsig_vardict:
+        input: config['project']['workpath']+"/vardict_out/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/vardict_out/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_vardict"
+        shell: "echo \"null\" > {output.genes}"
+
+  elif config['project']['annotation'] == "mm10":
+    rule mutsig_vardict:
+        input: config['project']['workpath']+"/vardict_out/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/vardict_out/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_vardict"
+        shell: "echo \"null\" > {output.genes}"
+
+else:
+    rule mutsig_vardict:
+        input: config['project']['workpath']+"/vardict_out/oncotator_out/final_filtered.maf",
+        output: genes=config['project']['workpath']+"/vardict_out/mutsigCV_out/somatic.sig_genes.txt",
+        params: rname="pl:mutsig_vardict"
+        shell: "echo \"null\" > {output.genes}"

Rules/vardict.rl

@@ -10,4 +10,4 @@ rule vardict:
                htmlstats=config['project']['workpath']+"/vardict_out/{x}.vardict.stats.html",
                out=config['project']['workpath']+"/vardict_out/{x}.snpeff.out"
        params: normalsample=lambda wildcards: config['project']['pairs'][wildcards.x][0],tumorsample=lambda wildcards: config['project']['pairs'][wildcards.x][1],targets="exome_targets.bed",knowns=config['references'][pfamily]['MUTECTVARIANTS'],mutect=config['bin'][pfamily]['MUTECT'],gatk=config['bin'][pfamily]['GATK'],genome=config['references'][pfamily]['MUTECTGENOME'],cosmic=config['references'][pfamily]['MUTECTCOSMIC'],snp=config['references'][pfamily]['MUTECTSNP'],snpeffgenome=config['references'][pfamily]['SNPEFF_GENOME'],snpeff=config['bin'][pfamily]['SNPEFF'],effconfig=config['references'][pfamily]['SNPEFF_CONFIG'],pon=config['references'][pfamily]['PON'],rname="pl:vardict"
-       shell:  "mkdir -p vardict_out; module load R/3.5; module load samtools/1.9; /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/vardict -G {params.genome} -f 0.05 -x 1000 -N {params.tumorsample} --nosv -b \"{input.tumor}|{input.normal}\" -y -t -Q 20 -c 1 -S 2 -E 3 {params.targets} | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/testsomatic.R | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/var2vcf_paired.pl -N {params.tumorsample} -f 0.05 > {output.vcf}; module load GATK/3.8-0; module load java/1.8.0_92; GATK -m 48G SelectVariants -R {params.genome} --variant {output.vcf} --excludeFiltered --discordance {params.pon} -o {output.filtvcf}; module load snpEff/4.3t; java -Xmx12g -jar $SNPEFF_JAR -v {params.snpeffgenome} -c {params.effconfig} -interval {params.targets} -cancer -canon -csvStats {output.csvstats} -stats {output.htmlstats} -cancerSamples pairs {output.filtvcf} > {output.out}"
+       shell:  "mkdir -p vardict_out; module load R/3.5; module load samtools/1.9; /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/vardict -G {params.genome} -f 0.05 -x 1000 -N {params.tumorsample} --nosv -b \"{input.tumor}|{input.normal}\" -t -Q 20 -c 1 -S 2 -E 3 {params.targets} | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/testsomatic.R | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/var2vcf_paired.pl -N {params.tumorsample} -f 0.05 > {output.vcf}; module load GATK/3.8-0; module load java/1.8.0_92; GATK -m 48G SelectVariants -R {params.genome} --variant {output.vcf} --excludeFiltered --discordance {params.pon} -o {output.filtvcf}; module load snpEff/4.3t; java -Xmx12g -jar $SNPEFF_JAR -v {params.snpeffgenome} -c {params.effconfig} -interval {params.targets} -cancer -canon -csvStats {output.csvstats} -stats {output.htmlstats} -cancerSamples pairs {output.filtvcf} > {output.out}"

Rules/vardict_tumoronly.rl

@@ -8,4 +8,4 @@ rule vardict_tumoronly:
                csvstats=config['project']['workpath']+"/vardict_out/{x}.vardict.stats.csv",
                htmlstats=config['project']['workpath']+"/vardict_out/{x}.vardict.stats.html",
        params: sample="{x}",pon=config['references'][pfamily]['PON'],targets="exome_targets.bed",knowns=config['references'][pfamily]['MUTECTVARIANTS'],mutect=config['bin'][pfamily]['MUTECT'],gatk=config['bin'][pfamily]['GATK'],genome=config['references'][pfamily]['MUTECTGENOME'],cosmic=config['references'][pfamily]['MUTECTCOSMIC'],snp=config['references'][pfamily]['MUTECTSNP'],snpeffgenome=config['references'][pfamily]['SNPEFF_GENOME'],snpeff=config['bin'][pfamily]['SNPEFF'],effconfig=config['references'][pfamily]['SNPEFF_CONFIG'],rname="pl:vardict"
-       shell:  "mkdir -p vardict_out; module load R/3.5; module load samtools/1.9; /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/vardict -G {params.genome} -f 0.05 -x 1000 --nosv -b {input.tumor} -y -t -Q 20 -c 1 -S 2 -E 3 {params.targets} | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/teststrandbias.R | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/var2vcf_valid.pl -N {params.sample} -S -E -f 0.05 > {output.vcf}; module load GATK/3.8-0; module load java/1.8.0_92; GATK -m 48G SelectVariants -R {params.genome} --variant {output.vcf} --discordance {params.pon} -o {output.filtvcf}; module load snpEff/4.3t; java -Xmx12g -jar $SNPEFF_JAR -v {params.snpeffgenome} -c {params.effconfig} -cancer -canon -csvStats {output.csvstats} -stats {output.htmlstats} -cancerSamples pairs {output.filtvcf} > {output.out}"
+       shell:  "mkdir -p vardict_out; module load R/3.5; module load samtools/1.9; /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/vardict -G {params.genome} -f 0.05 -x 1000 --nosv -b {input.tumor} -t -Q 20 -c 1 -S 2 -E 3 {params.targets} | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/teststrandbias.R | /data/CCBR_Pipeliner/db/PipeDB/bin/VarDict/var2vcf_valid.pl -N {params.sample} -S -E -f 0.05 > {output.vcf}; module load GATK/3.8-0; module load java/1.8.0_92; GATK -m 48G SelectVariants -R {params.genome} --variant {output.vcf} --discordance {params.pon} -o {output.filtvcf}; module load snpEff/4.3t; java -Xmx12g -jar $SNPEFF_JAR -v {params.snpeffgenome} -c {params.effconfig} -cancer -canon -csvStats {output.csvstats} -stats {output.htmlstats} -cancerSamples pairs {output.filtvcf} > {output.out}"