Merge pull request #318 from biowdl/shallow_mode

Add inputs for hfmtool's purple "shallow mode"

CHANGELOG.md

@@ -9,6 +9,8 @@ that users understand how the changes affect the new version.
 -->
 version 5.1.0-dev
 ---------------------------
++ Add the `highlyDiploidPercentage` and `somaticMinPuritySpread` inputs to the
+  hmtools PURPLE task.
 + Add a task for fastp.
 + Add a task for picard CollectInsertSizeMetrics.
 + Increased the timeMinutes runtime attribute for manta (somatic and germline) to `2880`.

fastp.wdl

@@ -38,6 +38,7 @@ task Fastp {
         Boolean performAdapterTrimming = true
         Boolean performQualityFiltering = true
         Boolean performLengthFiltering = true
+        Boolean? performPolyGTrimming
 
         Int threads = 4
         String memory = "50GiB"
@@ -50,6 +51,11 @@ task Fastp {
     String outputDirR1 = sub(outputPathR1, basename(outputPathR1), "")
     String outputDirR2 = sub(outputPathR2, basename(outputPathR2), "")
 
+    String polyGTrimmingFlag = if defined(performPolyGTrimming)
+        then
+            if select_first([performPolyGTrimming]) then "--trim_poly_g" else "--disable_trim_poly_g"
+        else ""
+
     Int? effectiveSplit = if select_first([split, 1]) > 1 then split else noneInt
 
     command <<<
@@ -77,7 +83,8 @@ task Fastp {
         ~{if defined(effectiveSplit) then "-d 0" else ""} \
         ~{if performAdapterTrimming then "" else "--disable_adapter_trimming"} \
         ~{if performQualityFiltering then "" else "--disable_quality_filtering"} \
-        ~{if performLengthFiltering then "" else "--disable_length_filtering"}
+        ~{if performLengthFiltering then "" else "--disable_length_filtering"} \
+        ~{polyGTrimmingFlag}
     >>>
 
     output {
@@ -108,6 +115,7 @@ task Fastp {
         performAdapterTrimming: {description: "Whether adapter trimming should be performed or not.", category: "advanced"}
         performQualityFiltering: {description: "Whether reads should be filtered based on quality scores.", category: "advanced"}
         performLengthFiltering: {description: "Whether reads shoulde be filtered based on lengths.", catgegory: "advanced"}
+        performPolyGTrimming: {description: "Whether or not poly-G-tail trimming should be performed. If undefined fastp's default behaviour will be used, ie. enabled for NextSeq/NovaSeq data as detected from read headers.", category: "advanced"}
         threads: {description: "The number of threads to use. Only used if the split input is not set.", category: "advanced"}
         memory: {description: "The amount of memory this job will use.", category: "advanced"}
         timeMinutes: {description: "The maximum amount of time the job will run in minutes.", category: "advanced"}

hmftools.wdl

@@ -1070,6 +1070,8 @@ task Purple {
         File driverGenePanel
         File somaticHotspots
         File germlineHotspots
+        Float? highlyDiploidPercentage
+        Float? somaticMinPuritySpread
         #The following should be in the same directory.
         File geneDataCsv
         File proteinFeaturesCsv
@@ -1103,6 +1105,8 @@ task Purple {
         -run_drivers \
         -somatic_hotspots ~{somaticHotspots} \
         -driver_gene_panel ~{driverGenePanel} \
+        ~{"-highly_diploid_percentage " + highlyDiploidPercentage} \
+        ~{"-somatic_min_purity_spread " + somaticMinPuritySpread} \
         -threads ~{threads}
     }
 
@@ -1182,6 +1186,8 @@ task Purple {
         driverGenePanel: {description: "A TSV file describing the driver gene panel.", category: "required"}
         somaticHotspots: {description: "A vcf file with hotspot somatic variant sites.", category: "required"}
         germlineHotspots: {description: "A vcf file with hotspot germline variant sites.", category: "required"}
+        highlyDiploidPercentage: {description: "Equivalent to PURPLE's `-highly_diploid_percentage` option.", category: "advanced"}
+        somaticMinPuritySpread: {description: "Equivalent to PURPLE's `-somatic_min_purity_spread` option.", category: "advanced"}
         geneDataCsv: {description: "A  CSV file containing gene information, must be in the same directory as `proteinFeaturesCsv`, `transExonDataCsv` and `transSpliceDataCsv`.", category: "required"}
         proteinFeaturesCsv: {description: "A  CSV file containing protein feature information, must be in the same directory as `geneDataCsv`, `transExonDataCsv` and `transSpliceDataCsv`.", category: "required"}
         transExonDataCsv: {description: "A  CSV file containing transcript exon information, must be in the same directory as `geneDataCsv`, `proteinFeaturesCsv` and `transSpliceDataCsv`.", category: "required"}