Add SNV annotation to test profile (#75)

genomic-medicine-sweden · Apr 17, 2024 · 3ebc4e8 · 3ebc4e8
1 parent 5313950
commit 3ebc4e8
Show file tree

Hide file tree

Showing 17 changed files with 278 additions and 23 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -9,6 +9,7 @@ Initial release of genomic-medicine-sweden/skierfe, created with the [nf-core](h
 
 ### `Added`
 
+- Added SNV annotation subworkflow to test profile [#75](https://github.com/genomic-medicine-sweden/skierfe/pull/75)
 - Refactored conditionally required parameters validation [#69](https://github.com/genomic-medicine-sweden/skierfe/pull/69)
 - Added CNV workflow to test profile [#68](https://github.com/genomic-medicine-sweden/skierfe/pull/68)
 - Updated and rename output directories [#65](https://github.com/genomic-medicine-sweden/skierfe/pull/65)

diff --git a/assets/test_data/snp_dbs.csv b/assets/test_data/snp_dbs.csv
@@ -1,2 +1,3 @@
 sample,file
 cadd,https://raw.githubusercontent.com/genomic-medicine-sweden/skierfe/dev/assets/test_data/cadd.v1.6.hg38.test_data.zip
+
diff --git a/conf/modules/prepare_genome.config b/conf/modules/prepare_genome.config
@@ -26,6 +26,10 @@ process {
     }
 
     withName: '.*:PREPARE_GENOME:MINIMAP2_INDEX_DIPCALL' {
-            ext.args = "-x asm5"
+        ext.args = "-x asm5"
+    }
+
+    withName: '.*:PREPARE_GENOME:UNTAR_VEP_CACHE' {
+        ext.when = { (params.vep_cache && params.vep_cache.endsWith("tar.gz")) }
     }
 }
diff --git a/conf/modules/snv_annotation.config b/conf/modules/snv_annotation.config
@@ -24,15 +24,12 @@ process {
         ]
     }
 
-    withName: '.*:SNV_ANNOTATION:BCFTOOLS_NORM' {
+    withName: '.*:SNV_ANNOTATION:BCFTOOLS_NORM_SINGLESAMPLE' {
         ext.args = "-m - --output-type b -w 10000"
     }
 
     withName: '.*:SNV_ANNOTATION:ECHTVAR_ANNO' {
 
-        // Do we want to filter variants here?
-        //ext.args = ""
-
         publishDir = [
             path: { "${params.outdir}/snv_annotation/echtvar/anno/${meta.id}" },
             mode: params.publish_dir_mode,
@@ -42,9 +39,6 @@ process {
 
     withName: '.*:SNV_ANNOTATION:ECHTVAR_ENCODE' {
 
-        // Do we want to filter variants here?
-        //ext.args = ""
-
         publishDir = [
             path: { "${params.outdir}/snv_annotation/echtvar/encode/${meta.id}" },
             mode: params.publish_dir_mode,

diff --git a/conf/test.config b/conf/test.config
@@ -39,8 +39,9 @@ params {
     // TRGT
     trgt_repeats = "https://raw.githubusercontent.com/genomic-medicine-sweden/skierfe/dev/assets/external/pathogenic_repeats.hg38.bed"
 
-    // Not tested
-    skip_snv_annotation = true
+    // SNV Annotation
+    vep_cache = "https://raw.githubusercontent.com/genomic-medicine-sweden/skierfe/dev/assets/test_data/vep_cache_test_data.tar.gz"
+    snp_db    = "https://raw.githubusercontent.com/genomic-medicine-sweden/skierfe/dev/assets/test_data/snp_dbs.csv"
 
     parallel_snv = 3 // Create 3 parallel DeepVariant processes
     preset = "revio"

diff --git a/modules.json b/modules.json
@@ -138,6 +138,11 @@
                         "branch": "master",
                         "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
                         "installed_by": ["modules"]
+                    },
+                    "untar": {
+                        "branch": "master",
+                        "git_sha": "5caf7640a9ef1d18d765d55339be751bb0969dfa",
+                        "installed_by": ["modules"]
                     }
                 }
             },

diff --git a/modules/nf-core/untar/environment.yml b/modules/nf-core/untar/environment.yml
diff --git a/modules/nf-core/untar/main.nf b/modules/nf-core/untar/main.nf
diff --git a/modules/nf-core/untar/meta.yml b/modules/nf-core/untar/meta.yml
diff --git a/modules/nf-core/untar/tests/main.nf.test b/modules/nf-core/untar/tests/main.nf.test
diff --git a/modules/nf-core/untar/tests/main.nf.test.snap b/modules/nf-core/untar/tests/main.nf.test.snap
diff --git a/modules/nf-core/untar/tests/tags.yml b/modules/nf-core/untar/tests/tags.yml
diff --git a/nextflow.config b/nextflow.config
@@ -32,6 +32,7 @@ params {
     trgt_repeats               = null
     snp_db                     = null
     vep_cache                  = null
+    vep_cache_version          = 110
 
     // HiFiCNV
     hificnv_xy                 = null

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -354,6 +354,11 @@
                     "default": 13,
                     "description": "Split SNV calling into n chunks",
                     "minimum": 1
+                },
+                "vep_cache_version": {
+                    "type": "integer",
+                    "default": 110,
+                    "description": "VEP cache version"
                 }
             },
             "required": ["preset"]
@@ -412,7 +417,7 @@
                 "vep_cache": {
                     "type": "string",
                     "description": "Path to directory of vep_cache",
-                    "format": "directory-path",
+                    "format": "path",
                     "exists": true
                 },
                 "bed": {

diff --git a/subworkflows/local/prepare_genome.nf b/subworkflows/local/prepare_genome.nf
@@ -1,11 +1,13 @@
-include { SAMTOOLS_FAIDX         } from '../../modules/nf-core/samtools/faidx/main'
-include { MINIMAP2_INDEX         } from '../../modules/nf-core/minimap2/index/main'
-include { GUNZIP as GUNZIP_FASTA } from '../../modules/nf-core/gunzip/main'
+include { SAMTOOLS_FAIDX           } from '../../modules/nf-core/samtools/faidx/main'
+include { MINIMAP2_INDEX           } from '../../modules/nf-core/minimap2/index/main'
+include { GUNZIP as GUNZIP_FASTA   } from '../../modules/nf-core/gunzip/main'
+include { UNTAR as UNTAR_VEP_CACHE } from '../../modules/nf-core/untar/main'
 
 workflow PREPARE_GENOME {
 
     take:
-    fasta_in // channel: [ val(meta), fasta ]
+    fasta_in     // channel: [ val(meta), fasta ]
+    ch_vep_cache // channel: [mandatory for annotation] [ path(cache) ]
 
     main:
     ch_versions = Channel.empty()
@@ -29,13 +31,22 @@ workflow PREPARE_GENOME {
     SAMTOOLS_FAIDX ( ch_fasta )
     MINIMAP2_INDEX ( ch_fasta )
 
+    UNTAR_VEP_CACHE (ch_vep_cache)
+
+    UNTAR_VEP_CACHE.out.untar
+        .map { meta, files -> [files] }
+        .collect()
+        .set { untarred_vep }
+
     // Gather versions
     ch_versions = ch_versions.mix(SAMTOOLS_FAIDX.out.versions.first())
     ch_versions = ch_versions.mix(MINIMAP2_INDEX.out.versions.first())
+    ch_versions = ch_versions.mix(UNTAR_VEP_CACHE.out.versions)
 
     emit:
-    fasta    = ch_fasta                           // channel: [ val(meta), fasta ]
-    fai      = SAMTOOLS_FAIDX.out.fai.collect()   // channel: [ val(meta), fai ]
-    mmi      = MINIMAP2_INDEX.out.index.collect() // channel: [ val(meta), mmi ]
-    versions = ch_versions                        // channel: [ versions.yml ]
+    fasta         = ch_fasta                           // channel: [ val(meta), fasta ]
+    fai           = SAMTOOLS_FAIDX.out.fai.collect()   // channel: [ val(meta), fai ]
+    vep_resources = untarred_vep                       // channel: [ path(cache) ]
+    mmi           = MINIMAP2_INDEX.out.index.collect() // channel: [ val(meta), mmi ]
+    versions      = ch_versions                        // channel: [ versions.yml ]
 }
diff --git a/subworkflows/local/snv_annotation.nf b/subworkflows/local/snv_annotation.nf
@@ -17,6 +17,7 @@ workflow SNV_ANNOTATION {
     ch_databases
     ch_fasta
     ch_vep_cache
+    val_vep_cache_version // string: [mandatory] default: 110
 
     main:
     ch_versions       = Channel.empty()
@@ -51,7 +52,7 @@ workflow SNV_ANNOTATION {
         vep_in,
         "GRCh38",
         "homo_sapiens",
-        "110",
+        val_vep_cache_version,
         ch_vep_cache,
         ch_fasta,
         []