Sentieon-haplotyper and joint-germline with Sentieon Genotyper and VQSR

.github/workflows/pytest-workflow.yml

@@ -56,6 +56,8 @@ jobs:
             tags: concatenate_vcfs
           - profile: "conda"
             tags: sentieon/bwamem
+          - profile: "singularity"
+            tags: concatenate_vcfs
           - profile: "singularity"
             tags: merge
           - profile: "singularity"

conf/modules/prepare_genome.config

@@ -76,7 +76,7 @@ process {
     }
 
     withName: 'TABIX_DBSNP' {
-        ext.when         = { !params.dbsnp_tbi && params.dbsnp && ((params.step == "mapping" || params.step == "markduplicates" || params.step == "prepare_recalibration") || params.tools && (params.tools.split(',').contains('controlfreec') || params.tools.split(',').contains('haplotypecaller') || params.tools.split(',').contains('mutect2'))) }
+        ext.when         = { !params.dbsnp_tbi && params.dbsnp && ((params.step == "mapping" || params.step == "markduplicates" || params.step == "prepare_recalibration") || params.tools && (params.tools.split(',').contains('controlfreec') || params.tools.split(',').contains('haplotypecaller') || params.tools.split(',').contains('sentieon_haplotyper') || params.tools.split(',').contains('mutect2'))) }
         publishDir       = [
             enabled: (params.save_reference || params.build_only_index),
             mode: params.publish_dir_mode,
@@ -96,7 +96,7 @@ process {
     }
 
     withName: 'TABIX_KNOWN_INDELS' {
-        ext.when         = { !params.known_indels_tbi && params.known_indels && (params.step == 'mapping' || params.step == "markduplicates" || params.step == 'prepare_recalibration' || (params.tools && params.tools.split(',').contains('haplotypecaller')) ) }
+        ext.when         = { !params.known_indels_tbi && params.known_indels && (params.step == 'mapping' || params.step == "markduplicates" || params.step == 'prepare_recalibration' || (params.tools && (params.tools.split(',').contains('haplotypecaller') || params.tools.split(',').contains('sentieon_haplotyper'))) ) }
         publishDir       = [
             enabled: (params.save_reference || params.build_only_index),
             mode: params.publish_dir_mode,
@@ -106,7 +106,7 @@ process {
     }
 
     withName: 'TABIX_KNOWN_SNPS' {
-        ext.when         = { !params.known_snps_tbi && params.known_snps && (params.step == 'mapping' || params.step == "markduplicates" || params.step == 'prepare_recalibration' || (params.tools && params.tools.split(',').contains('haplotypecaller')) ) }
+        ext.when         = { !params.known_snps_tbi && params.known_snps && (params.step == 'mapping' || params.step == "markduplicates" || params.step == 'prepare_recalibration' || (params.tools && (params.tools.split(',').contains('haplotypecaller') || params.tools.split(',').contains('sentieon_haplotyper'))) ) }
         publishDir       = [
             enabled: (params.save_reference || params.build_only_index),
             mode: params.publish_dir_mode,

conf/modules/sentieon_haplotyper.config

@@ -0,0 +1,85 @@
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Config file for defining DSL2 per module options and publishing paths
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Available keys to override module options:
+        ext.args   = Additional arguments appended to command in module.
+        ext.args2  = Second set of arguments appended to command in module (multi-tool modules).
+        ext.args3  = Third set of arguments appended to command in module (multi-tool modules).
+        ext.prefix = File name prefix for output files.
+        ext.when   = When to run the module.
+----------------------------------------------------------------------------------------
+*/
+
+// SENTIEON HAPLOTYPER
+
+process {
+
+    withName: 'SENTIEON_HAPLOTYPER' {
+        // ext.args         = { params.joint_germline ? "-ERC GVCF" : "" }  // TO-DO: Change that
+        // ext.prefix       = { meta.num_intervals <= 1 ? ( params.joint_germline ? "${meta.id}.haplotyper.g" : "${meta.id}.haplotyper" ) : ( params.joint_germline ? "${meta.id}.haplotyper.${intervals.simpleName}.g" :"${meta.id}.haplotyper.${intervals.simpleName}" ) }  // old value
+        ext.prefix       = { meta.num_intervals <= 1 ? "${meta.id}.haplotyper" : "${meta.id}.haplotyper.${intervals.simpleName}" }
+        ext.when         = { params.tools && params.tools.split(',').contains('sentieon_haplotyper') }
+        publishDir       = [
+            enabled: !params.joint_germline,
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/variant_calling/"},
+            pattern: "*{vcf.gz,vcf.gz.tbi}",
+            saveAs: { meta.num_intervals > 1 ? null : "sentieon_haplotyper/${meta.id}/${it}" }
+        ]
+    }
+
+    withName: 'MERGE_SENTIEON_HAPLOTYPER_VCFS' {
+        ext.prefix       = { params.joint_germline ? "${meta.id}.haplotyper.g" : "${meta.id}.haplotyper.unfiltered" }  // TO-DO:  This shouldn't be relevant for gvcf. But check with the joint-germline flow.
+        publishDir       = [
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/variant_calling/sentieon_haplotyper/${meta.id}/" },
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+
+    withName: 'MERGE_SENTIEON_HAPLOTYPER_GVCFS' {
+        ext.prefix       = { "${meta.id}.haplotyper.g" }
+        publishDir       = [
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/variant_calling/sentieon_haplotyper/${meta.id}/" },
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: '.*BAM_VARIANT_CALLING_SENTIEON_HAPLOTYPER:VCF_VARIANT_FILTERING_GATK:FILTERVARIANTTRANCHES' {
+        ext.prefix       = {"${meta.id}.haplotyper"}
+        ext.args         = { "--info-key CNN_1D" }
+        publishDir       = [
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/variant_calling/sentieon_haplotyper/${meta.id}/"},
+            pattern: "*{vcf.gz,vcf.gz.tbi}"
+        ]
+    }
+
+    if (params.tools && params.tools.split(',').contains('haplotypecaller')) {
+        withName: 'NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_GERMLINE_ALL:BAM_VARIANT_CALLING_HAPLOTYPECALLER:BAM_MERGE_INDEX_SAMTOOLS:(MERGE_BAM|INDEX_MERGE_BAM)' {
+            ext.prefix       = { "${meta.id}.realigned" }
+            publishDir       = [
+                enabled: true,
+                mode: params.publish_dir_mode,
+                path: { "${params.outdir}/variant_calling/haplotypecaller/${meta.id}/" },
+                pattern: "*{bam,bai}"
+            ]
+        }
+    }
+
+    if (params.tools && params.tools.split(',').contains('sentieon_haplotyper')) {
+        withName: 'NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_GERMLINE_ALL:BAM_VARIANT_CALLING_SENTIEON_HAPLOTYPER:BAM_MERGE_INDEX_SAMTOOLS:(MERGE_BAM|INDEX_MERGE_BAM)' {
+            ext.prefix       = { "${meta.id}.realigned" }
+            publishDir       = [
+                enabled: true,
+                mode: params.publish_dir_mode,
+                path: { "${params.outdir}/variant_calling/sentieon_haplotyper/${meta.id}/" },
+                pattern: "*{bam,bai}"
+            ]
+        }
+    }
+
+}

conf/modules/sentieon_joint_germline.config

@@ -0,0 +1,84 @@
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Config file for defining DSL2 per module options and publishing paths
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Available keys to override module options:
+        ext.args   = Additional arguments appended to command in module.
+        ext.args2  = Second set of arguments appended to command in module (multi-tool modules).
+        ext.args3  = Third set of arguments appended to command in module (multi-tool modules).
+        ext.prefix = File name prefix for output files.
+        ext.when   = When to run the module.
+----------------------------------------------------------------------------------------
+*/
+
+// SENTIEON+GATK4 JOINT_GERMLINE  // TO-DO: Figure out how much of this config is needed or wanted.
+
+process {
+
+    withName: 'NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_GERMLINE_ALL:BAM_JOINT_CALLING_GERMLINE_SENTIEON:GATK4_GENOMICSDBIMPORT' {
+        ext.args         = { "--genomicsdb-shared-posixfs-optimizations true --bypass-feature-reader "}
+        ext.prefix       = { "${meta.intervals_name}.joint" }
+    }
+
+    withName: 'NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_GERMLINE_ALL:BAM_JOINT_CALLING_GERMLINE_SENTIEON:SENTIEON_GVCFTYPER' {
+        ext.args = { "--allow-old-rms-mapping-quality-annotation-data" }
+        ext.prefix       = { meta.intervals_name }
+    }
+
+    if (params.tools && params.tools.contains('sentieon_haplotyper') && params.joint_germline) {
+        withName: 'NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_GERMLINE_ALL:BAM_JOINT_CALLING_GERMLINE_SENTIEON::BCFTOOLS_SORT' {
+            ext.prefix       = { "${vcf.baseName.minus("vcf")}sort" }
+            publishDir       = [
+                enabled: false
+            ]
+        }
+    }
+
+    withName: 'NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_GERMLINE_ALL:BAM_JOINT_CALLING_GERMLINE_SENTIEON:MERGE_GENOTYPEGVCFS' {
+        ext.prefix       = "joint_germline"
+        publishDir       = [
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/variant_calling/sentieon_haplotyper/joint_variant_calling/" },
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
+            pattern: "*{vcf.gz,vcf.gz.tbi}"
+        ]
+    }
+
+    withName: 'NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_GERMLINE_ALL:BAM_JOINT_CALLING_GERMLINE_SENTIEON:MERGE_VQSR' {
+        ext.prefix       = "joint_germline_recalibrated"
+        publishDir       = [
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/variant_calling/sentieon_haplotyper/joint_variant_calling/"},
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
+            pattern: "*{vcf.gz,vcf.gz.tbi}"
+        ]
+    }
+
+    withName: 'SENTIEON_VARCAL_INDEL' {
+        ext.prefix = { "${meta.id}_INDEL" }
+        ext.args = "--annotation QD --annotation MQRankSum --annotation ReadPosRankSum --annotation FS --annotation SOR --annotation DP --var_type INDEL"
+        publishDir = [
+            enabled: false
+        ]
+    }
+
+    withName: 'SENTIEON_APPLYVARCAL_INDEL' {
+        ext.prefix    = { "${meta.id}_INDEL" }
+        ext.args      = '--sensitivity 99.9 --var_type INDEL'
+    }
+
+    withName: 'SENTIEON_VARCAL_SNP' {
+        ext.prefix = { "${meta.id}_SNP" }
+        ext.args = "--annotation QD --annotation MQRankSum --annotation ReadPosRankSum --annotation FS --annotation SOR --annotation DP --var_type SNP"
+        publishDir = [
+            enabled: false
+        ]
+    }
+
+    withName: 'SENTIEON_APPLYVARCAL_SNP' {
+        ext.prefix    = { "${meta.id}_SNP" }
+        ext.args      = '--sensitivity 99.9 --var_type SNP'
+    }
+
+
+}

modules.json

@@ -371,14 +371,34 @@
                         "git_sha": "371eff7748d769c2ddc8bd593773523a364a52fe",
                         "installed_by": ["modules"]
                     },
+                    "sentieon/applyvarcal": {
+                        "branch": "master",
+                        "git_sha": "49a64392c75566394617a372d0fdd52caf754cfc",
+                        "installed_by": ["modules"]
+                    },
                     "sentieon/bwamem": {
                         "branch": "master",
-                        "git_sha": "8c159ebb28cfa08890042a133bfab78de932eff7",
+                        "git_sha": "b9829e1064382745d8dff7f1d74d2138d2864f71",
                         "installed_by": ["modules"]
                     },
                     "sentieon/dedup": {
                         "branch": "master",
-                        "git_sha": "8c04ecc0c5d4ad102755b919b2525481f0380090",
+                        "git_sha": "b9829e1064382745d8dff7f1d74d2138d2864f71",
+                        "installed_by": ["modules"]
+                    },
+                    "sentieon/gvcftyper": {
+                        "branch": "master",
+                        "git_sha": "aa591ada04e79a771d242cd967648e6a1ae06ccf",
+                        "installed_by": ["modules"]
+                    },
+                    "sentieon/haplotyper": {
+                        "branch": "master",
+                        "git_sha": "b9829e1064382745d8dff7f1d74d2138d2864f71",
+                        "installed_by": ["modules"]
+                    },
+                    "sentieon/varcal": {
+                        "branch": "master",
+                        "git_sha": "641b422042e9d401d00f3e1a4f7ae59d53c9bfe0",
                         "installed_by": ["modules"]
                     },
                     "snpeff/download": {