ExpansionHunter extract sex from a ped file. (#281)

* ExpansionHunter extract sex from a ped file. * Update wdl/ExpansionHunter.wdl Co-authored-by: Mark Walker <[email protected]> * Update wdl/ExpansionHunterScatter.wdl Co-authored-by: Mark Walker <[email protected]> * Update wdl/ExpansionHunter.wdl Co-authored-by: Mark Walker <[email protected]> * Update wdl/ExpansionHunterScatter.wdl Co-authored-by: Mark Walker <[email protected]> * Update wdl/ExpansionHunter.wdl Co-authored-by: Mark Walker <[email protected]> * Add parameter metadata for few arguments. * Errout if sample ID or ped file are not defined & few bug fixes. * Improve on determining sample sex. * Remove output_prefix & make sample_id required. * Update wdl/ExpansionHunter.wdl Co-authored-by: Mark Walker <[email protected]> * Update wdl/ExpansionHunterScatter.wdl Co-authored-by: Mark Walker <[email protected]> * Update a comment. Co-authored-by: Mark Walker <[email protected]>
broadinstitute · Jan 11, 2022 · 479d2ac · 479d2ac
1 parent 6437798
commit 479d2ac
Show file tree

Hide file tree

Showing 2 changed files with 37 additions and 28 deletions.
diff --git a/wdl/ExpansionHunter.wdl b/wdl/ExpansionHunter.wdl
@@ -20,11 +20,17 @@ workflow ExpansionHunter {
         File reference_fasta
         File? reference_fasta_index
         File variant_catalog
-        File? output_prefix
+        String sample_id
+        File? ped_file
         String expansion_hunter_docker
         RuntimeAttr? runtime_attr
     }
 
+    parameter_meta {
+        ped_file: "This file is used to extract the sex of the bam_or_cram file."
+        sample_id: "The ped_file needs to be provided as well to determine sample sex. The ID must match the sample ID given in the second column (`Individual ID` column) of the given PED file. This ID will also be used as an output prefix."
+    }
+
     Boolean is_bam = basename(bam_or_cram, ".bam") + ".bam" == basename(bam_or_cram)
     File bam_or_cram_index_ =
         if defined(bam_or_cram_index) then
@@ -36,23 +42,15 @@ workflow ExpansionHunter {
         reference_fasta_index,
         reference_fasta + ".fai"])
 
-    String output_prefix_ =
-        if defined(output_prefix) then
-            select_first([output_prefix])
-        else
-            if is_bam then
-                basename(bam_or_cram, ".bam")
-            else
-                basename(bam_or_cram, ".cram")
-
     call RunExpansionHunter {
         input:
             bam_or_cram = bam_or_cram,
             bam_or_cram_index = bam_or_cram_index_,
             reference_fasta = reference_fasta,
             reference_fasta_index = reference_fasta_index_,
             variant_catalog = variant_catalog,
-            output_prefix = output_prefix_,
+            sample_id = sample_id,
+            ped_file = ped_file,
             expansion_hunter_docker = expansion_hunter_docker,
             runtime_attr_override = runtime_attr,
     }
@@ -72,28 +70,39 @@ task RunExpansionHunter {
         File reference_fasta
         File reference_fasta_index
         File variant_catalog
-        String output_prefix
+        String sample_id
+        File? ped_file
         String expansion_hunter_docker
         RuntimeAttr? runtime_attr_override
     }
 
     output {
-        File json = "${output_prefix}.json"
-        File vcf = "${output_prefix}.vcf"
-        File overlapping_reads = "${output_prefix}_realigned.bam"
-        File timing = "${output_prefix}_timing.tsv"
+        File json = "${sample_id}.json"
+        File vcf = "${sample_id}.vcf"
+        File overlapping_reads = "${sample_id}_realigned.bam"
+        File timing = "${sample_id}_timing.tsv"
     }
 
     command <<<
         set -euxo pipefail
 
+        sex=""
+        if ~{defined(ped_file)}; then
+            sex=$(awk -F '\t' '{if ($2 == "~{sample_id}") {if ($5 == "1") {print "--sex male"; exit 0} else if ($5 == "2") {print "--sex female"; exit 0}}}' < ~{ped_file} )
+            if [ "$sex" = "" ]; then
+                echo "The Sex of the sample defined in the PED file is other than male or female. ExpansionHunter only supports male or female samples."
+                exit 1
+            fi
+        fi
+
         ExpansionHunter \
             --reads ~{bam_or_cram} \
             --reference ~{reference_fasta} \
             --variant-catalog ~{variant_catalog} \
-            --output-prefix ~{output_prefix} \
+            --output-prefix ~{sample_id} \
             --cache-mates \
-            --record-timing
+            --record-timing \
+            $sex
     >>>
 
     RuntimeAttr runtime_attr_str_profile_default = object {

diff --git a/wdl/ExpansionHunterScatter.wdl b/wdl/ExpansionHunterScatter.wdl
@@ -8,14 +8,20 @@ workflow ExpansionHunterScatter {
     input {
         Array[File] bams_or_crams
         Array[File]? bams_or_crams_indexes
-        Array[String]? sample_ids
+        Array[String] sample_ids
+        File? ped_file
         File reference_fasta
         File? reference_fasta_index
         File variant_catalog
         String expansion_hunter_docker
         RuntimeAttr? runtime_attr
     }
 
+    parameter_meta {
+        ped_file: "This file is used to extract the sex of the BAM/CRAM files."
+        sample_ids: "One ID per sample, in the same order as the files in bams_or_crams. These IDs must match the ID given in the second column (`Individual ID` column) of the given PED file. These IDs will also be used as an output prefix."
+    }
+
     scatter (i in range(length(bams_or_crams))) {
         File bam_or_cram_ = bams_or_crams[i]
         Boolean is_bam =
@@ -28,14 +34,7 @@ workflow ExpansionHunterScatter {
         File reference_fasta_index_ = select_first([
             reference_fasta_index, reference_fasta + ".fai"])
 
-        String output_prefix =
-            if defined(sample_ids) then
-                select_first([sample_ids])[i]
-            else
-                if is_bam then
-                    basename(bam_or_cram_, ".bam")
-                else
-                    basename(bam_or_cram_, ".cram")
+        String sample_id = sample_ids[i]
 
         call ExpansionHunter.ExpansionHunter as expanionHunter {
             input:
@@ -44,7 +43,8 @@ workflow ExpansionHunterScatter {
                 reference_fasta=reference_fasta,
                 reference_fasta_index=reference_fasta_index_,
                 variant_catalog=variant_catalog,
-                output_prefix=output_prefix,
+                sample_id=sample_id,
+                ped_file=ped_file,
                 expansion_hunter_docker=expansion_hunter_docker,
                 runtime_attr=runtime_attr
         }