fixed bugs in expanded WES tumor only pipeline; changed admixture out…

…puts to all be printed in admixture_out; incorporated vardict somatic calling
CCBR · May 26, 2019 · f00d7cb · f00d7cb
1 parent 9b15248
commit f00d7cb
Show file tree

Hide file tree

Showing 14 changed files with 33 additions and 32 deletions.
diff --git a/Rules/admixture_germline.rl b/Rules/admixture_germline.rl
@@ -4,7 +4,7 @@ rule admixture_germline:
             mergedvcf=temp("admixture_out/samples_and_knowns.vcf"),
             ped=temp("admixture_out/samples_and_knowns_filtered.ped"),
             map=temp("admixture_out/samples_and_knowns_filtered.map"),
-            recodeped="admixture_out/samples_and_knowns_filtered_recode.ped",
+            recodeped=temp("admixture_out/samples_and_knowns_filtered_recode.ped"),
             admix="admixture_out/samples_and_knowns_filtered_recode.Q",
             table="admixture_out/admixture_table.tsv",
             plot="admixture_out/admixture_mqc.png"

diff --git a/Rules/admixture_somatic.rl b/Rules/admixture_somatic.rl
@@ -4,7 +4,7 @@ rule admixture_somatic:
             mergedvcf=temp("admixture_out/samples_and_knowns.vcf"),
             ped=temp("admixture_out/samples_and_knowns_filtered.ped"),
             map=temp("admixture_out/samples_and_knowns_filtered.map"),
-            recodeped="admixture_out/samples_and_knowns_filtered_recode.ped",
+            recodeped=temp("admixture_out/samples_and_knowns_filtered_recode.ped"),
             admix="admixture_out/samples_and_knowns_filtered_recode.Q",
             table="admixture_out/admixture_table.tsv",
             plot="admixture_out/admixture_mqc.png"

diff --git a/Rules/all-exomeseq-germline.rl b/Rules/all-exomeseq-germline.rl
@@ -13,6 +13,6 @@ rule all_exomeseq_germline:
     output: 
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
 
             """
diff --git a/Rules/all-exomeseq-somatic-tumoronly.rl b/Rules/all-exomeseq-somatic-tumoronly.rl
@@ -18,7 +18,7 @@ if config['project']['annotation'] == "hg19":
             config['project']['workpath']+"/vardict_out/mutsigCV_out/somatic.sig_genes.txt",
             config['project']['workpath']+"/merged_somatic_variants/mutsigCV_out/somatic.sig_genes.txt",
             expand(config['project']['workpath']+"/mutect_out/{s}"+".FINAL.vcf",s=samples),
-            expand(config['project']['workpath']+"/vardict_out/{s}"+".vcf",s=samples),
+            expand(config['project']['workpath']+"/vardict_out/{s}"+".FINAL.vcf",s=samples),
             expand(config['project']['workpath']+"/vardict_out/oncotator_out/{s}"+".maf",s=samples),
             expand(config['project']['workpath']+"/mutect_out/oncotator_out/{s}"+".maf",s=samples),
             expand(config['project']['workpath']+"/merged_somatic_variants/{s}"+".merged.vcf",s=samples),
@@ -30,7 +30,7 @@ if config['project']['annotation'] == "hg19":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 
@@ -55,7 +55,7 @@ elif config['project']['annotation'] == "hg38":
             config['project']['workpath']+"/vardict_out/mutsigCV_out/somatic.sig_genes.txt",
             config['project']['workpath']+"/merged_somatic_variants/mutsigCV_out/somatic.sig_genes.txt",
             expand(config['project']['workpath']+"/mutect_out/{s}"+".FINAL.vcf",s=samples),
-            expand(config['project']['workpath']+"/vardict_out/{s}"+".vcf",s=samples),
+            expand(config['project']['workpath']+"/vardict_out/{s}"+".FINAL.vcf",s=samples),
             expand(config['project']['workpath']+"/vardict_out/oncotator_out/{s}"+".maf",s=samples),
             expand(config['project']['workpath']+"/mutect_out/oncotator_out/{s}"+".maf",s=samples),
             expand(config['project']['workpath']+"/merged_somatic_variants/{s}"+".merged.vcf",s=samples),
@@ -65,7 +65,7 @@ elif config['project']['annotation'] == "hg38":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 
@@ -90,7 +90,7 @@ elif config['project']['annotation'] == "mm10":
             config['project']['workpath']+"/vardict_out/mutsigCV_out/somatic.sig_genes.txt",
             config['project']['workpath']+"/merged_somatic_variants/mutsigCV_out/somatic.sig_genes.txt",
             expand(config['project']['workpath']+"/mutect_out/{s}"+".FINAL.vcf",s=samples),
-            expand(config['project']['workpath']+"/vardict_out/{s}"+".vcf",s=samples),
+            expand(config['project']['workpath']+"/vardict_out/{s}"+".FINAL.vcf",s=samples),
             expand(config['project']['workpath']+"/vardict_out/oncotator_out/{s}"+".maf",s=samples),
             expand(config['project']['workpath']+"/mutect_out/oncotator_out/{s}"+".maf",s=samples),
             expand(config['project']['workpath']+"/merged_somatic_variants/{s}"+".merged.vcf",s=samples),
@@ -100,6 +100,6 @@ elif config['project']['annotation'] == "mm10":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
diff --git a/Rules/all-exomeseq-somatic.rl b/Rules/all-exomeseq-somatic.rl
@@ -32,7 +32,7 @@ if config['project']['annotation'] == "hg19":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 elif config['project']['annotation'] == "hg38":
@@ -69,7 +69,7 @@ elif config['project']['annotation'] == "hg38":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 
@@ -101,6 +101,6 @@ elif config['project']['annotation'] == "mm10":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
diff --git a/Rules/all-rnaseqvargerm.rl b/Rules/all-rnaseqvargerm.rl
@@ -14,6 +14,6 @@ rule all_rnaseqvargerm:
     output: 
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; rm *sorted.bam.bai; mv *sorted.txt logfiles/; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; mv *.fin.bam.intervals logfiles/; rm *realign.bai; rm *sorted.bam.bai; mv *sorted.txt logfiles/; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
 
             """
diff --git a/Rules/all-wgs-somatic-tumoronly.rl b/Rules/all-wgs-somatic-tumoronly.rl
@@ -18,7 +18,7 @@ if config['project']['annotation'] == "hg19":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.6; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 elif config['project']['annotation'] == "hg38":
@@ -41,7 +41,7 @@ elif config['project']['annotation'] == "hg38":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.6; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 
@@ -62,6 +62,6 @@ elif config['project']['annotation'] == "mm10":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.6; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
diff --git a/Rules/all-wgs-somatic.rl b/Rules/all-wgs-somatic.rl
@@ -34,7 +34,7 @@ if config['project']['annotation'] == "hg19":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.6; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 elif config['project']['annotation'] == "hg38":
@@ -74,7 +74,7 @@ elif config['project']['annotation'] == "hg38":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.6; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
 
@@ -112,6 +112,6 @@ elif config['project']['annotation'] == "mm10":
     output:
     params: rname="final"
     shell:  """
-             module load multiqc/1.6; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
+             module load multiqc/1.7; multiqc -f .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped logfiles/
 
             """
diff --git a/Rules/all-wgslow.rl b/Rules/all-wgslow.rl
@@ -25,7 +25,7 @@ if config['project']['annotation'] == "hg19":
     output: 
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
 
             """
 
@@ -56,7 +56,7 @@ elif config['project']['annotation'] == "hg38":
     output: 
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
 
             """
 
@@ -85,6 +85,6 @@ elif config['project']['annotation'] == "mm10":
     output: 
     params: rname="final"
     shell:  """
-             module load multiqc/1.4; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
+             module load multiqc/1.7; multiqc -f -e featureCounts .; mv *.out slurmfiles/; perl Scripts/summarize_usage.pl; rm *realign.bai; mv distance.cluster0 distance.cluster1 distance.cluster2 distance.cluster3 distance.nosex samples.txt plink.map plink.ped *.avia_status.txt *.avia.log *_genotypes.vcf logfiles/
 
             """
diff --git a/Rules/maftools_tumoronly.rl b/Rules/maftools_tumoronly.rl
@@ -5,7 +5,7 @@ rule maftools_mutect2_tumoronly:
           summary=config['project']['workpath']+"/mutect2_out/mutect2_maf_summary.pdf",
           oncoprint=config['project']['workpath']+"/mutect2_out/mutect2_oncoplot.pdf",
   params: dir=config['project']['workpath'],rname="pl:maftools"
-  shell: "cat mutect2_out/oncotator_out/*.maf > mutect2_out/oncotator_out/mutect2_variants.maf; perl Scripts/prep_mafs.pl mutect2_out/oncotator_out/mutect2_variants.maf mutect2; module load R/3.5; Rscript Scripts/maftools.R {params.dir}/mutect2_out/oncotator_out/ final_filtered.maf {params.dir}/mutect2_out/mutect2_maf_summary.pdf {output.mutect2oncoprint}"
+  shell: "cat mutect2_out/oncotator_out/*.maf > mutect2_out/oncotator_out/mutect2_variants.maf; perl Scripts/prep_mafs.pl mutect2_out/oncotator_out/mutect2_variants.maf mutect2; module load R/3.5; Rscript Scripts/maftools.R {params.dir}/mutect2_out/oncotator_out/ final_filtered.maf {params.dir}/mutect2_out/mutect2_maf_summary.pdf {output.oncoprint}"
 
 rule maftools_tumoronly_mutect:
   input: mutectmaf=expand(config['project']['workpath']+"/mutect_out/oncotator_out/{s}.maf",s=samples),
@@ -14,7 +14,7 @@ rule maftools_tumoronly_mutect:
           summary=config['project']['workpath']+"/mutect_out/mutect_maf_summary.pdf",
           oncoprint=config['project']['workpath']+"/mutect_out/mutect_oncoplot.pdf",
   params: dir=config['project']['workpath'],rname="pl:maftools"
-  shell: "cat mutect_out/oncotator_out/*.maf > mutect_out/oncotator_out/mutect_variants.maf; perl Scripts/prep_mafs.pl mutect_out/oncotator_out/mutect_variants.maf mutect; module load R/3.5; Rscript Scripts/maftools.R {params.dir}/mutect_out/oncotator_out/ final_filtered.maf {params.dir}/mutect_out/mutect_maf_summary.pdf {output.mutectoncoprint}"
+  shell: "cat mutect_out/oncotator_out/*.maf > mutect_out/oncotator_out/mutect_variants.maf; perl Scripts/prep_mafs.pl mutect_out/oncotator_out/mutect_variants.maf mutect; module load R/3.5; Rscript Scripts/maftools.R {params.dir}/mutect_out/oncotator_out/ final_filtered.maf {params.dir}/mutect_out/mutect_maf_summary.pdf {output.oncoprint}"
 
 rule maftools_tumoronly_vardict:
   input: vardictmaf=expand(config['project']['workpath']+"/vardict_out/oncotator_out/{s}.maf",s=samples),

diff --git a/Rules/mkdir_somatic_tumoronly.rl b/Rules/mkdir_somatic_tumoronly.rl
@@ -7,4 +7,4 @@ rule mkdir_somatic:
                 mut2chrom_dir=config['project']['workpath']+"/mutect2_out/chrom_files",
                 manta_dir=config['project']['workpath']+"/manta_out"
         params: rname="pl:mkdir"
-        shell:  "mkdir -p germline_vcfs; mkdir -p mutect2_out; mkdir -p mutect2_out/oncotator_out; mkdir -p mutect2_out/mutsigCV_out; mkdir -p mutect2_out/chrom_files; mkdir mutect_out; mkdir mutect_out/oncotator_out; mkdir_mutect_out/mutsigCV_out; mkdir vardict_out; mkdir vardict_out/oncotator_out; mkdir vardict_out/mutsigCV_out; mkdir merged_somatic_variants; mkdir merged_somatic_variants/oncotator_out; mkdir merged_somatic_variants/mutsigCV_out; mkdir -p manta_out"
+        shell:  "mkdir -p germline_vcfs; mkdir -p mutect2_out; mkdir -p mutect2_out/oncotator_out; mkdir -p mutect2_out/mutsigCV_out; mkdir -p mutect2_out/chrom_files; mkdir -p mutect_out; mkdir -p mutect_out/oncotator_out; mkdir -p mutect_out/mutsigCV_out; mkdir -p vardict_out; mkdir -p vardict_out/oncotator_out; mkdir -p vardict_out/mutsigCV_out; mkdir -p merged_somatic_variants; mkdir -p merged_somatic_variants/oncotator_out; mkdir -p merged_somatic_variants/mutsigCV_out; mkdir -p manta_out"