modernize WDL

mlin · Aug 4, 2020 · 0ac63a1 · 0ac63a1
1 parent 606ba80
commit 0ac63a1
Show file tree

Hide file tree

Showing 2 changed files with 81 additions and 51 deletions.
diff --git a/spvcf.wdl b/spvcf.wdl
@@ -1,33 +1,40 @@
+version 1.0
+
 task spvcf_encode {
-    File vcf_gz
-    Boolean multithread = false
-    String release = "v1.0.0"
+    input {
+        File vcf_gz
+        Boolean multithread = false
+        String release = "v1.1.0"
+        Int cpu = if multithread then 8 else 4
+    }
 
     parameter_meta {
         vcf_gz: "stream"
     }
 
-    command {
-        set -ex -o pipefail
+    command <<<
+        set -euxo pipefail
 
-        apt-get update -qq && apt-get install -y -qq pigz wget
-        wget -nv https://github.com/mlin/spVCF/releases/download/${release}/spvcf
-        wget -nv https://github.com/dnanexus-rnd/GLnexus/raw/master/cli/dxapplet/resources/usr/local/bin/bgzip
-        chmod +x spvcf bgzip
+        apt-get -qq update && apt-get install -y wget tabix
+        wget -nv https://github.com/mlin/spVCF/releases/download/~{release}/spvcf
+        chmod +x spvcf
 
         threads_arg=""
-        if [ "${multithread}" == "true" ]; then
-            threads_arg="--threads $(nproc)"
+        if [ "~{multithread}" == "true" ]; then
+            threads_arg="--threads 4"
         fi
 
-        nm=$(basename "${vcf_gz}" .vcf.gz)
-        nm="$nm.spvcf.gz"
+        nm=$(basename "~{vcf_gz}" .vcf.gz)
+        nm="${nm}.spvcf.gz"
         mkdir out
-        pigz -dc "${vcf_gz}" | ./spvcf encode $threads_arg | ./bgzip -@ $(nproc) > "out/$nm"
-    }
+        bgzip -dc "~{vcf_gz}" | ./spvcf encode $threads_arg | bgzip -@ 4 > "out/${nm}"
+    >>>
 
     runtime {
-        docker: "ubuntu:18.04"
+        docker: "ubuntu:20.04"
+        cpu: cpu
+        memory: "~{cpu} GB"
+        disks: "local-disk ~{ceil(size(vcf_gz,'GB'))} SSD"
     }
 
     output {
@@ -36,29 +43,33 @@ task spvcf_encode {
 }
 
 task spvcf_decode {
-    File spvcf_gz
-    String release = "v1.0.0"
+    input {
+        File spvcf_gz
+        String release = "v1.1.0"
+    }
 
     parameter_meta {
         spvcf_gz: "stream"
     }
 
-    command {
-        set -ex -o pipefail
+    command <<<
+        set -euxo pipefail
 
-        apt-get update -qq && apt-get install -y -qq pigz wget
-        wget -nv https://github.com/mlin/spVCF/releases/download/${release}/spvcf
-        wget -nv https://github.com/dnanexus-rnd/GLnexus/raw/master/cli/dxapplet/resources/usr/local/bin/bgzip
-        chmod +x spvcf bgzip
+        apt-get -qq update && apt-get install -y wget tabix
+        wget -nv https://github.com/mlin/spVCF/releases/download/~{release}/spvcf
+        chmod +x spvcf
 
-        nm=$(basename "${spvcf_gz}" .spvcf.gz)
-        nm="$nm.vcf.gz"
+        nm=$(basename "~{spvcf_gz}" .spvcf.gz)
+        nm="${nm}.vcf.gz"
         mkdir out
-        pigz -dc "${spvcf_gz}" | ./spvcf decode | ./bgzip -@ $(nproc) > "out/$nm"
-    }
+        bgzip -dc "~{spvcf_gz}" | ./spvcf decode | bgzip -@ 4 > "out/${nm}"
+    >>>
 
     runtime {
-        docker: "ubuntu:18.04"
+        docker: "ubuntu:20.04"
+        cpu: 4
+        memory: "4 GB"
+        disks: "local-disk ~{10*ceil(size(spvcf_gz,'GB'))} SSD"
     }
 
     output {
@@ -67,35 +78,40 @@ task spvcf_decode {
 }
 
 task spvcf_squeeze {
-    File vcf_gz
-    Boolean multithread = false
-    String release = "v1.0.0"
+    input {
+        File vcf_gz
+        Boolean multithread = false
+        String release = "v1.1.0"
+        Int cpu = if multithread then 8 else 4
+    }
 
     parameter_meta {
         vcf_gz: "stream"
     }
 
-    command {
-        set -ex -o pipefail
+    command <<<
+        set -euxo pipefail
 
-        apt-get update -qq && apt-get install -y -qq pigz wget
-        wget -nv https://github.com/mlin/spVCF/releases/download/${release}/spvcf
-        wget -nv https://github.com/dnanexus-rnd/GLnexus/raw/master/cli/dxapplet/resources/usr/local/bin/bgzip
-        chmod +x spvcf bgzip
+        apt-get -qq update && apt-get install -y wget tabix
+        wget -nv https://github.com/mlin/spVCF/releases/download/~{release}/spvcf
+        chmod +x spvcf
 
         threads_arg=""
-        if [ "${multithread}" == "true" ]; then
-            threads_arg="--threads $(nproc)"
+        if [ "~{multithread}" == "true" ]; then
+            threads_arg="--threads 4"
         fi
 
-        nm=$(basename "${vcf_gz}" .vcf.gz)
-        nm="$nm.squeeze.vcf.gz"
+        nm=$(basename "~{vcf_gz}" .vcf.gz)
+        nm="${nm}.squeeze.vcf.gz"
         mkdir out
-        pigz -dc "${vcf_gz}" | ./spvcf squeeze $threads_arg | ./bgzip -@ $(nproc) > "out/$nm"
-    }
+        bgzip -dc "~{vcf_gz}" | ./spvcf squeeze $threads_arg | bgzip -@ 4 > "out/${nm}"
+    >>>
 
     runtime {
-        docker: "ubuntu:18.04"
+        docker: "ubuntu:20.04"
+        cpu: cpu
+        memory: "~{cpu} GB"
+        disks: "local-disk ~{ceil(size(vcf_gz,'GB'))} SSD"
     }
 
     output {

diff --git a/test/test_spvcf.wdl b/test/test_spvcf.wdl
@@ -1,7 +1,11 @@
+version 1.0
+
 import "spvcf.wdl" as tasks
 
 workflow test_spvcf {
-    File vcf_gz # pVCF
+    input {
+        File vcf_gz # pVCF
+    }
 
     # spVCF-encode the pVCF
     call tasks.spvcf_encode {
@@ -37,11 +41,21 @@ workflow test_spvcf {
 }
 
 task verify_identical_gz_content {
-    File gz1
-    File gz2
+    input {
+        File gz1
+        File gz2
+    }
 
-    command {
-        set -e -o pipefail
-        cmp --silent <(gzip -dc "${gz1}") <(gzip -dc "${gz2}")
+    command <<<
+        set -euxo pipefail
+        apt-get -qq update && apt-get install -y tabix
+        cmp --silent <(bgzip -dc "~{gz1}") <(bgzip -dc "~{gz2}")
+    >>>
+
+    runtime {
+        docker: "ubuntu:20.04"
+        cpu: 4
+        memory: "4 GB"
+        disks: "local-disk ~{ceil(size(gz2,'GB')+size(gz1,'GiB'))+4} SSD"
     }
 }