From 847ad71a26b3a1ddc1fc06c2fda349fc620ad2b5 Mon Sep 17 00:00:00 2001
From: Ruben Vorderman <r.h.p.vorderman@lumc.nl>
Date: Wed, 9 Apr 2025 19:40:40 +0200
Subject: [PATCH 1/2] Update vt to allow a filter expression and compressed
 indexed output

---
 CHANGELOG.md |  1 +
 vt.wdl       | 28 ++++++++++++++++++++++------
 2 files changed, 23 insertions(+), 6 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 96adc8fa..dfa40b75 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,6 +10,7 @@ that users understand how the changes affect the new version.
 
 version 6.0.0-dev
 ---------------------------
++ Update vt task to allow a filter expression and compress and index the output.
 + MultiQC image updated to version 1.28
 + Samtools merge now has options added for merging RG and PG headers.
 + Samtools merge default thread count increased based on the number of files.
diff --git a/vt.wdl b/vt.wdl
index 4da2d8cd..4ced1d2a 100644
--- a/vt.wdl
+++ b/vt.wdl
@@ -27,27 +27,39 @@ task Normalize {
         File referenceFasta
         File referenceFastaFai
         Boolean ignoreMaskedRef = false
-        String outputPath = "./vt/normalized_decomposed.vcf"
+        String outputPath = "./vt/normalized_decomposed.vcf.gz"
+        String? filterExpression
+
+        Int compressionLevel = 1
 
         String memory = "4GiB"
-        Int timeMinutes = 30
-        String dockerImage = "quay.io/biocontainers/vt:0.57721--hdf88d34_2"
+        Int timeMinutes = 10 + ceil(size(inputVCF, "GiB") * 240)
+        String dockerImage = "quay.io/biocontainers/vt:0.57721--h2419454_12"
     }
 
     command {
         set -eo pipefail
         mkdir -p "$(dirname ~{outputPath})"
-        vt normalize ~{inputVCF} \
+        vt view -h \
+        ~{"-f " + filterExpression} \
+        ~{inputVCF} \
+        | vt normalize - \
         -r ~{referenceFasta} \
         ~{true="-m " false="" ignoreMaskedRef} \
-        | vt decompose -s - -o ~{outputPath}
+        | vt decompose -s - \
+        | vt view - \
+        -c ~{compressionLevel} \
+        -o ~{outputPath} 
+        vt index ~{outputPath}
     }
 
     output {
         File outputVcf = outputPath
+        File outputVcfIndex = outputPath + ".tbi"
     }
 
     runtime {
+        cpu: 1
         memory: memory
         time_minutes: timeMinutes
         docker: dockerImage
@@ -61,11 +73,15 @@ task Normalize {
         referenceFastaFai: {description: "The index for the reference fasta file.", category: "required"}
         ignoreMaskedRef: {description: "Warns but does not exit when REF is inconsistent with masked reference sequence for non SNPs.", category: "advanced"}
         outputPath: {description: "The location the output VCF file should be written.", category: "common"}
+        filterExpression: {description: "See https://genome.sph.umich.edu/wiki/Vt#Filters for valid expressions.", category: "common"}
+        compressionLevel: {description: "Compression level for the out vcf.gz file.", category: "advanced"}
+
         memory: {description: "The memory required to run the programs.", category: "advanced"}
         timeMinutes: {description: "The maximum amount of time the job will run in minutes.", category: "advanced"}
         dockerImage: {description: "The docker image used for this task. Changing this may result in errors which the developers may choose not to address.", category: "advanced"}
 
         # outputs
-        outputVcf: {description: "Normalized & decomposed VCF file."}
+        outputVcf: {description: "Normalized and decomposed VCF file."}
+        outputVcfIndex: {description: "Index for normalized and decomposed VCF file."}
     }
 }

From 57018dd55c43af0013f48a61e5119128ccd87d3f Mon Sep 17 00:00:00 2001
From: Ruben Vorderman <r.h.p.vorderman@lumc.nl>
Date: Fri, 11 Apr 2025 11:28:56 +0200
Subject: [PATCH 2/2] Properly quote vt filter

---
 vt.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/vt.wdl b/vt.wdl
index 4ced1d2a..635641e9 100644
--- a/vt.wdl
+++ b/vt.wdl
@@ -41,7 +41,7 @@ task Normalize {
         set -eo pipefail
         mkdir -p "$(dirname ~{outputPath})"
         vt view -h \
-        ~{"-f " + filterExpression} \
+        ~{"-f '" + filterExpression}~{true="'" false="" defined(filterExpression)} \
         ~{inputVCF} \
         | vt normalize - \
         -r ~{referenceFasta} \