Add XY filtration workflow

config/default.config

@@ -20,10 +20,12 @@ params {
     picard_version = "2.26.10"
     pipeval_version = "4.0.0-rc.2"
     gatkfilter_version = "v1.0.0"
+    hail_version = "branch-mmootor-fix-spark-permission"
     docker_image_gatk = "broadinstitute/gatk:${params.gatk_version}"
     docker_image_picard = "${-> params.docker_container_registry}/picard:${params.picard_version}"
     docker_image_pipeval = "${-> params.docker_container_registry}/pipeval:${params.pipeval_version}"
     docker_image_gatkfilter = "${-> params.docker_container_registry}/gatk:${params.gatkfilter_version}"
+    docker_image_hail = "${-> params.docker_container_registry}/hail:${params.hail_version}"
 
     emit_all_confident_sites = false
 }
@@ -36,7 +38,7 @@ process {
     cache = true
 
     executor = 'local'
-    
+
     // Other directives or options that should apply for every process
 
     // total amount of resources avaible to the pipeline

config/schema.yaml

@@ -3,10 +3,18 @@ patient_id:
   type: 'String'
   required: true
   help: 'Patient ID'
+sample_sex:
+  type: 'String'
+  required: true
+  help: 'Sample Sex'
 dataset_id:
   type: 'String'
   required: true
   help: 'Dataset ID'
+genome_build:
+  type: 'String'
+  required: true
+  help: 'Genome build, GRCh37 or GRCh38'
 output_dir:
   type: 'Path'
   mode: 'w'
@@ -62,6 +70,11 @@ bundle_phase1_1000g_snps_high_conf_vcf_gz:
   mode: 'r'
   required: true
   help: 'Absolute path to high-confidence 1000g SNPs VCF'
+par_bed:
+  type: 'Path'
+  mode: 'r'
+  required: true
+  help: 'Absolute path to Pseudo-autosomal Region (PAR) BED'
 base_resource_update:
   type: 'ResourceUpdateNamespace'
   required: false

config/template.config

@@ -11,6 +11,11 @@ params {
     dataset_id = ''
     blcds_registered_dataset = false // if you want the output to be registered
 
+    genome_build = "GRCh38"
+
+    // Input patient sex if known - male or female. Leave empty if not known.
+    sample_sex = ''
+
     output_dir = '/path/to/output/directory'
 
     // Set to false to disable the publish rule and delete intermediate files as they're no longer needed
@@ -43,6 +48,9 @@ params {
     bundle_omni_1000g_2p5_vcf_gz = "/hot/resource/tool-specific-input/GATK/GRCh38/1000G_omni2.5.hg38.vcf.gz"
     bundle_phase1_1000g_snps_high_conf_vcf_gz = "/hot/resource/tool-specific-input/GATK/GRCh38/1000G_phase1.snps.high_confidence.hg38.vcf.gz"
 
+    // Specify BED file path for Pseudoautosomal Region (PAR)
+    par_bed = ""
+
     // Base resource allocation updater
     // See README for adding parameters to update the base resource allocations
 }

main.nf

@@ -68,6 +68,7 @@ include {
     } from './module/merge-vcf.nf'
 include { recalibrate_variants } from './module/workflow-recalibrate-variants.nf'
 include { filter_gSNP_GATK } from './module/filter-gsnp.nf'
+include { filter_XY } from './module/filter-xy.nf'
 include { calculate_sha512 } from './module/checksum.nf'
 
 // Returns the index file for the given bam or vcf
@@ -104,6 +105,12 @@ workflow {
         }
         .set{ input_ch_collected_files }
 
+    script_dir_ch = Channel.fromPath(
+        "$projectDir/script",
+        checkIfExists: true
+        )
+        .collect()
+
     /**
     *   Input validation
     */
@@ -248,6 +255,20 @@ workflow {
         recalibrate_variants.out.output_ch_recalibrated_variants
     )
 
+    filter_xy_ch = recalibrate_variants.out.output_ch_recalibrated_variants
+        .map { it -> [it[0], it[1], it[2]] }
+
+    script_dir_ch = Channel.fromPath(
+        "$projectDir/script",
+        checkIfExists: true
+        )
+        .collect()
+
+    filter_XY(
+        filter_xy_ch,
+        params.par_bed,
+        script_dir_ch
+        )
     /**
     *   Calculate checksums for output files
     */

module/filter-xy.nf

@@ -0,0 +1,69 @@
+include { generate_standard_filename; sanitize_string } from '../external/pipeline-Nextflow-module/modules/common/generate_standardized_filename/main.nf'
+
+/*
+    Nextflow module for filtering chrX and chrY variant calls based on sample sex
+
+    input:
+        sample_id: identifier for sample
+        sample_vcf: path to VCF to filter
+        sample_vcf_tbi: path to index of VCF to filter
+
+    params:
+        params.output_dir_base: string(path)
+        params.log_output_dir: string(path)
+        params.docker_image_hail: string
+        params.sample_sex: string
+        params.par_bed: string(path)
+*/
+
+process filter_XY {
+    container params.docker_image_hail
+
+    publishDir path: "${params.output_dir_base}/output",
+      mode: "copy",
+      pattern: '*.vcf.bgz*',
+      saveAs: {
+        "${output_filename}_${sanitize_string(file(it).getName().replace("${sample_id}_", ""))}"
+        }
+
+    publishDir path: "${params.log_output_dir}/process-log",
+      pattern: ".command.*",
+      mode: "copy",
+      saveAs: {
+        "${task.process.replace(':', '/')}/${task.process.split(':')[-1]}-${sample_id}-${interval_id}/log${file(it).getName()}"
+        }
+
+    input:
+    tuple val(sample_id), path(recalibrated_vcf), path(recalibrated_vcf_tbi)
+    path(par_bed)
+    path(script_dir)
+
+    output:
+    path(".command.*")
+    path("${output_filename}_XY_filtered.vcf.bgz")
+    path("${output_filename}_XY_filtered.vcf.bgz.tbi")
+
+    script:
+    output_filename = generate_standard_filename(
+        "Hail-${params.hail_version}",
+        params.dataset_id,
+        sample_id,
+        [additional_tools:["GATK-${params.gatk_version}"]]
+        )
+    """
+    set -euo pipefail
+
+    zgrep "##source=" ${recalibrated_vcf} > ./vcf_source.txt
+
+    cat ./vcf_source.txt
+
+    python ${script_dir}/filter_xy_call.py \
+        --sample_name ${output_filename} \
+        --input_vcf ${recalibrated_vcf} \
+        --vcf_source_file ./vcf_source.txt \
+        --sample_sex ${params.sample_sex} \
+        --par_bed ${par_bed} \
+        --genome_build ${params.genome_build} \
+        --output_dir .
+    """
+}

script/filter_xy_call.py

@@ -0,0 +1,156 @@
+#!/usr/bin/env python3
+"""
+Filter XY calls from call-gSNP single sample VCF file
+
+Steps:
+- Extract autosomes and chrX/Y variants from input VCF
+- Filter chrX/Y variants
+- Merge autosomal and filtered chrX/Y variants
+
+Filter criteria:
+- Extract XY calls
+- Extract XY calls overlapping with Pseudo-Autosomal Regions (PARs)
+- For non-PAR
+    - Male sample:
+        - Filter out heterozygous GT calls in chrX and chrY
+        - Transform homozygous GT=1/1 to hemizygous GT=1
+    - Female sample: Filter out chrY calls
+
+Dependencies:
+- Python 3
+- HAIL python library (pip install hail)
+
+Note:
+- Do not export VCF to a path that is being read from in the same pipeline,\
+based on HAIL recommendation
+"""
+
+import os
+import argparse
+import hail as hl
+
+script_dir = os.getcwd()
+
+parser = argparse.ArgumentParser()
+parser.add_argument(
+    '--sample_name',
+    dest='sample_name',
+    help = 'Sample name',
+    required=True
+    )
+parser.add_argument(
+    '--input_vcf',
+    dest='input_vcf',
+    help = 'Input single sample VCF file path',
+    required=True
+    )
+parser.add_argument(
+    '--vcf_source_file',
+    dest='vcf_source_file',
+    help = 'A TXT file containing variant caller source details (eg. ##source=HaplotypeCaller)',
+    required=True
+    )
+parser.add_argument(
+    '--sample_sex',
+    dest='sample_sex',
+    help = 'Sample sex, XY or XX',
+    required=True
+    )
+parser.add_argument(
+    '--par_bed',
+    dest='par_bed',
+    help = 'Input BED file path for Pseudo-Autosomal Regions (PAR)',
+    required=True
+    )
+parser.add_argument(
+    '--genome_build',
+    dest='genome_build',
+    help = 'Genome build of input VCF, GRCh37 or GRCh38',
+    required=True
+    )
+parser.add_argument(
+    '--output_dir',
+    dest='output_dir',
+    help = 'Output path where filtered XY variant VCF will be written',
+    required=True
+    )
+
+args = parser.parse_args()
+
+sample_name = args.sample_name
+sample_sex = args.sample_sex
+vcf_file = args.input_vcf
+vcf_source_file = args.vcf_source_file
+par_bed = args.par_bed
+genome_build = args.genome_build
+output_dir = args.output_dir
+
+#Import PAR BED file
+par = hl.import_bed(
+    path = par_bed,
+    reference_genome = genome_build,
+    skip_invalid_intervals = True
+    )
+
+#Extract VCF file header
+vcf_header = hl.get_vcf_metadata(vcf_file)
+
+#Import VCF file into a hail MatrixTable
+vcf_matrix = hl.import_vcf(
+    path = vcf_file,
+    reference_genome = genome_build,
+    force_bgz = True
+    )
+
+#Filter XY calls
+##Extract XY calls
+X_contig = vcf_matrix.locus.contig.startswith('chrX') | vcf_matrix.locus.contig.startswith('X')
+Y_contig = vcf_matrix.locus.contig.startswith('chrY') | vcf_matrix.locus.contig.startswith('Y')
+extract_condition = (X_contig) | (Y_contig)
+vcf_XY = vcf_matrix.filter_rows(extract_condition)
+print('chrX/Y variants before XY filtration:', vcf_XY.count())
+
+##Extract autosomes
+vcf_autosomes = vcf_matrix.filter_rows(~extract_condition)
+
+##Extract PAR and non-PAR regions
+par_variants = vcf_XY.filter_rows(hl.is_defined(par[vcf_XY.locus]))
+non_par_variants = vcf_XY.filter_rows(hl.is_missing(par[vcf_XY.locus]))
+
+if sample_sex == 'XY':
+    #If MALE (XY), remove heterozygous non-PAR chrX calls
+    non_par_filtered_variants = non_par_variants.filter_rows(
+        hl.agg.all(
+            non_par_variants.GT.is_diploid() & non_par_variants.GT.is_hom_var()
+            )
+        )
+    non_par_filtered_variants = non_par_filtered_variants.annotate_entries(
+        GT = hl.call(non_par_filtered_variants.GT[0])
+        )
+
+elif sample_sex == 'XX':
+    #If Female (XX), remove non-PAR chrY calls
+    non_par_filtered_variants = non_par_variants.filter_rows(
+        non_par_variants.locus.contig.startswith('chrX') | \
+            non_par_variants.locus.contig.startswith('X')
+        )
+
+#Combine PAR and filtered non-PAR regions
+par_non_par = [par_variants, non_par_filtered_variants]
+filterXY = hl.MatrixTable.union_rows(*par_non_par)
+print('chrX/Y variant counts after XY filtration:', filterXY.count())
+
+#Combine filtered X/Y + autosomal variants
+autosomes_XYfiltered = [vcf_autosomes, filterXY]
+output_vcf = hl.MatrixTable.union_rows(*autosomes_XYfiltered)
+
+#Export MatrixTable to VCF
+output_file = output_dir + '/' + sample_name + '_XY_filtered.vcf.bgz'
+
+hl.export_vcf(
+    dataset = output_vcf,
+    output = output_file,
+    tabix = True,
+    metadata = vcf_header,
+    append_to_header = vcf_source_file
+    )