Moving up ability to set mismatch quality related filter thresholds t…

…o the input YAML

definitions/detect_variants_nonhuman.wdl

@@ -63,6 +63,9 @@ workflow detectVariantsNonhuman {
     Array[String] variants_to_table_fields = ["CHROM", "POS", "ID", "REF", "ALT", "set", "AC", "AF"]
     Array[String] variants_to_table_genotype_fields = ["GT", "AD"]
     Array[String] vep_to_table_fields = []
+
+    Int? max_mm_qualsum_diff
+    Int? max_var_mm_qualsum    
   }
 
   call m.mutect {
@@ -77,7 +80,9 @@ workflow detectVariantsNonhuman {
     normal_bam_bai=normal_bam_bai,
     interval_list=roi_intervals,
     scatter_count=scatter_count,
-    fp_min_var_freq=fp_min_var_freq
+    fp_min_var_freq=fp_min_var_freq,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum
   }
 
   call sapp.strelkaAndPostProcessing as strelka {
@@ -94,7 +99,9 @@ workflow detectVariantsNonhuman {
     cpu_reserved=strelka_cpu_reserved,
     normal_sample_name=normal_sample_name,
     tumor_sample_name=tumor_sample_name,
-    fp_min_var_freq=fp_min_var_freq
+    fp_min_var_freq=fp_min_var_freq,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum    
   }
 
   call vpapp.varscanPreAndPostProcessing as varscan {
@@ -114,7 +121,9 @@ workflow detectVariantsNonhuman {
     p_value=varscan_p_value,
     max_normal_freq=varscan_max_normal_freq,
     normal_sample_name=normal_sample_name,
-    tumor_sample_name=tumor_sample_name
+    tumor_sample_name=tumor_sample_name,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum    
   }
 
   call cv.combineVariants as combine {

definitions/somatic_exome_nonhuman.wdl

@@ -71,6 +71,9 @@ workflow somaticExomeNonhuman {
 
     String tumor_sample_name
     String normal_sample_name
+
+    Int? max_mm_qualsum_diff
+    Int? max_var_mm_qualsum    
   }
 
   call aen.alignmentExomeNonhuman as tumorAlignmentAndQc {
@@ -161,7 +164,9 @@ workflow somaticExomeNonhuman {
     vep_to_table_fields=vep_to_table_fields,
     tumor_sample_name=tumor_sample_name,
     normal_sample_name=normal_sample_name,
-    strelka_exome_mode=true
+    strelka_exome_mode=true,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum    
   }
 
   call btc.bamToCram as tumorBamToCram {

definitions/subworkflows/fp_filter.wdl

@@ -20,6 +20,8 @@ workflow fpFilter {
     String variant_caller
     String? sample_name
     Float? fp_min_var_freq
+    Int? max_mm_qualsum_diff
+    Int? max_var_mm_qualsum    
   }
 
   call vs.vcfSanitize as sanitizeVcf {
@@ -54,6 +56,8 @@ workflow fpFilter {
     vcf=index.indexed_vcf,
     sample_name=sample_name,
     fp_min_var_freq=fp_min_var_freq,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum,
     output_vcf_basename = variant_caller + "_full"
   }
 

definitions/subworkflows/mutect.wdl

@@ -23,6 +23,9 @@ workflow mutect {
     Int scatter_count
     String tumor_sample_name
     Float? fp_min_var_freq
+
+    Int? max_mm_qualsum_diff
+    Int? max_var_mm_qualsum    
   }
 
   call sil.splitIntervalList {
@@ -66,7 +69,9 @@ workflow mutect {
     vcf_tbi=indexVcf.indexed_vcf_tbi,
     variant_caller="mutect",
     sample_name=tumor_sample_name,
-    fp_min_var_freq=fp_min_var_freq
+    fp_min_var_freq=fp_min_var_freq,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum
   }
 
   output {

definitions/subworkflows/strelka_and_post_processing.wdl

@@ -31,6 +31,10 @@ workflow strelkaAndPostProcessing {
     File? call_regions_tbi
 
     Float? fp_min_var_freq
+
+    Int? max_mm_qualsum_diff
+    Int? max_var_mm_qualsum
+
   }
 
   call s.strelka {
@@ -99,7 +103,9 @@ workflow strelkaAndPostProcessing {
     vcf_tbi=regionFilter.filtered_vcf_tbi,
     sample_name=tumor_sample_name,
     variant_caller="strelka",
-    fp_min_var_freq=fp_min_var_freq
+    fp_min_var_freq=fp_min_var_freq,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum    
   }
 
   output {

definitions/subworkflows/varscan_pre_and_post_processing.wdl

@@ -32,6 +32,10 @@ workflow varscanPreAndPostProcessing {
     Float? p_value
     Float? max_normal_freq
     Int scatter_count = 50
+
+    Int? max_mm_qualsum_diff
+    Int? max_var_mm_qualsum
+
   }
 
   call sil.splitIntervalList {
@@ -156,6 +160,8 @@ workflow varscanPreAndPostProcessing {
     vcf_tbi=index.indexed_vcf_tbi,
     fp_min_var_freq=varscan_min_var_freq,
     sample_name=tumor_sample_name,
+    max_mm_qualsum_diff=max_mm_qualsum_diff,
+    max_var_mm_qualsum=max_var_mm_qualsum,    
     variant_caller="varscan"
   }
 

definitions/tools/fp_filter.wdl

@@ -12,6 +12,8 @@ task fpFilter {
     String output_vcf_basename = "fpfilter"
     String sample_name = "TUMOR"
     Float fp_min_var_freq = 0.05
+    Int? max_mm_qualsum_diff
+    Int? max_var_mm_qualsum
   }
 
   Int space_needed_gb = 10 + round(size(vcf, "GB")*2 + size([reference, reference_fai, reference_dict, bam], "GB"))
@@ -26,7 +28,17 @@ task fpFilter {
 
   String output_vcf = output_vcf_basename + ".vcf"
   command <<<
-    /usr/bin/perl /usr/bin/fpfilter.pl --bam-readcount /usr/bin/bam-readcount --samtools /opt/samtools/bin/samtools --output ~{output_vcf} --reference ~{reference} --bam-file ~{bam} --vcf-file ~{vcf} --sample ~{sample_name} --min-var-freq ~{fp_min_var_freq}
+    /usr/bin/perl /usr/bin/fpfilter.pl \
+    --bam-readcount /usr/bin/bam-readcount \
+    --samtools /opt/samtools/bin/samtools \
+    --output ~{output_vcf} \
+    --reference ~{reference} \
+    --bam-file ~{bam} \
+    --vcf-file ~{vcf} \
+    --sample ~{sample_name} \
+    --min-var-freq ~{fp_min_var_freq} \
+    ~{if defined(max_mm_qualsum_diff) then "--max-mm-qualsum-diff ~{max_mm_qualsum_diff}" else ""} \
+    ~{if defined(max_var_mm_qualsum) then "--max_var_mm_qualsum ~{max_var_mm_qualsum}" else ""}
   >>>
 
   output {