add bcftools csq as an annotation tool

Author: priesgo

Makefile

@@ -28,3 +28,4 @@ test:
 	bash test/scripts/run_test_10.sh
 	bash test/scripts/run_test_11.sh
 	bash test/scripts/run_test_12.sh
+	bash test/scripts/run_test_13.sh

main.nf

@@ -6,7 +6,7 @@ include { FILTER_VCF } from './modules/01_filter'
 include { BCFTOOLS_NORM; VT_DECOMPOSE_COMPLEX; REMOVE_DUPLICATES } from './modules/02_normalization'
 include { SUMMARY_VCF; SUMMARY_VCF as SUMMARY_VCF_2 } from './modules/03_summary'
 include { VAFATOR; MULTIALLELIC_FILTER } from './modules/04_vafator'
-include { VARIANT_ANNOTATION } from './modules/05_variant_annotation'
+include { VARIANT_ANNOTATION_SNPEFF; VARIANT_ANNOTATION_BCFTOOLS } from './modules/05_variant_annotation'
 
 params.help= false
 params.input_vcfs = false
@@ -19,6 +19,8 @@ params.input_clonalities = false
 
 
 params.reference = false
+params.gff = false
+
 params.output = "output"
 params.skip_normalization = false
 params.skip_decompose_complex = false
@@ -123,12 +125,6 @@ workflow {
             .join(input_purities.groupTuple(), remainder: true)
             .join(input_clonalities.groupTuple(), remainder: true)
 
-        //if (params.input_purities) {
-        //    vafator_input = vafator_input.join(input_purities.groupTuple())
-        //}
-        //if (params.input_clonalities) {
-        //    vafator_input = vafator_input.join(input_clonalities.groupTuple())
-        //}
         VAFATOR(vafator_input)
 
         final_vcfs = VAFATOR.out.annotated_vcf
@@ -139,8 +135,12 @@ workflow {
     }
 
     if (params.snpeff_organism) {
-        VARIANT_ANNOTATION(final_vcfs)
-        final_vcfs = VARIANT_ANNOTATION.out.annotated_vcf
+        VARIANT_ANNOTATION_SNPEFF(final_vcfs)
+        final_vcfs = VARIANT_ANNOTATION_SNPEFF.out.annotated_vcf
+    }
+    else if (params.gff) {
+        VARIANT_ANNOTATION_BCFTOOLS(final_vcfs)
+        final_vcfs = VARIANT_ANNOTATION_BCFTOOLS.out.annotated_vcf
     }
 }
 

modules/05_variant_annotation.nf

@@ -6,7 +6,7 @@ params.snpeff_organism = false
 params.snpeff_args = ""
 
 
-process VARIANT_ANNOTATION {
+process VARIANT_ANNOTATION_SNPEFF {
     cpus params.cpus
     memory params.memory
     publishDir "${params.output}/${name}", mode: "copy"
@@ -25,3 +25,25 @@ process VARIANT_ANNOTATION {
     snpEff eff ${datadir_arg} ${params.snpeff_args} -nodownload ${params.snpeff_organism} ${vcf} > ${name}.annotated.vcf
     """
 }
+
+process VARIANT_ANNOTATION_BCFTOOLS {
+    cpus params.cpus
+    memory params.memory
+    publishDir "${params.output}/${name}", mode: "copy"
+
+    conda (params.enable_conda ? "conda-forge::libgcc-ng=10.3.0 conda-forge::gsl=2.7 bioconda::bcftools=1.15.1" : null)
+
+    input:
+        tuple val(name), file(vcf)
+
+    output:
+    tuple val(name), file("${name}.annotated.vcf") , emit: annotated_vcf
+
+    """
+    bcftools csq \\
+        --fasta-ref ${params.reference} \\
+        --gff-annot ${params.gff} ${vcf} \\
+        --output-type v \\
+        --output ${name}.annotated.vcf
+    """
+}

test/data/Homo_sapiens.GRCh37.87.minimal.gff3.gz

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+### this test cannot be automated as it relies on SnpEff references which need to be downloaded beforehand
+source test/scripts/assert.sh
+output_folder=test/output/test13
+data_folder=`pwd`/test/data
+
+nextflow main.nf -profile test,conda --output $output_folder \
+  --gff ${data_folder}/Homo_sapiens.GRCh37.87.minimal.gff3.gz \
+  --reference ${data_folder}/ucsc.hg19.minimal.fasta
+
+test -s $output_folder/single_sample/single_sample.annotated.vcf || { echo "Missing test 1 output file!"; exit 1; }
+test -s $output_folder/tumor_normal/tumor_normal.annotated.vcf || { echo "Missing test 1 output file!"; exit 1; }
+
+assert_eq `grep -v '#' $output_folder/single_sample/single_sample.annotated.vcf| grep BCSQ | wc -l | cut -d' ' -f 1` 32 "Wrong number of variants"
+assert_eq `grep -v '#' $output_folder/tumor_normal/tumor_normal.annotated.vcf | grep BCSQ | wc -l | cut -d' ' -f 1` 32 "Wrong number of variants"

test/scripts/run_test_13.sh

@@ -2,7 +2,7 @@
 
 ### this test cannot be automated as it relies on SnpEff references which need to be downloaded beforehand
 source test/scripts/assert.sh
-output_folder=test/output/test8
+output_folder=test/output/test9
 snpeff_datadir=/home/you/snpeff
 
 nextflow main.nf -profile test,conda --output $output_folder --snpeff_organism hg19 --snpeff_datadir $snpeff_datadir