Merge pull request #521 from genomic-medicine-sweden/vcf2cytosurerename

Add option to annotate vcf2cytosure file with different(customer) ids
nf-core · Mar 1, 2024 · 4f02eb0 · 4f02eb0
2 parents 653623c + 4f088f3
commit 4f02eb0
Show file tree

Hide file tree

Showing 6 changed files with 104 additions and 15 deletions.
diff --git a/assets/sample_id_map.json b/assets/sample_id_map.json
@@ -0,0 +1,24 @@
+{
+    "$schema": "http://json-schema.org/draft-07/schema",
+    "$id": "https://raw.githubusercontent.com/nf-core/raredisease/master/assets/mobile_element_references_schema.json",
+    "title": "Schema for customerid_internalid_map",
+    "description": "Schema for the file provided with params.customerid_internalid_map",
+    "type": "array",
+    "items": {
+        "type": "object",
+        "properties": {
+            "customer_id": {
+                "type": "string",
+                "pattern": "^\\S+$",
+                "errorMessage": "Customer id cannot contain spaces"
+            },
+            "internal_id": {
+                "type": "string",
+                "pattern": "^\\S+$",
+                "errorMessage": "Internal id cannot contain spaces",
+                "meta": ["id"]
+            }
+        },
+        "required": ["customer_id", "internal_id"]
+    }
+}
diff --git a/conf/modules/generate_cytosure_files.config b/conf/modules/generate_cytosure_files.config
@@ -30,7 +30,14 @@ process {
                 '--output-type z',
                 "--exclude 'gnomad_svAF > 0.05'"
                 ].join(' ') }
-            ext.prefix = { ${meta.id} }
+            ext.prefix = { "${meta.id}" }
+        }
+
+        withName: '.*GENERATE_CYTOSURE_FILES:BCFTOOLS_REHEADER_SV_VCF' {
+            beforeScript = { "echo ${meta.custid} > ${meta.custid}.txt" }
+            ext.args   = { "--samples ${meta.custid}.txt" }
+            ext.args2  = "--output-type v"
+            ext.prefix = { "${meta.custid}" }
         }
 
         withName: '.*GENERATE_CYTOSURE_FILES:VCF2CYTOSURE' {
@@ -39,7 +46,7 @@ process {
                 '--size 5000',
                 '--maxbnd 5000'
                 ].join(' ') }
-            ext.prefix = { "${meta.id}" }
+            ext.prefix = { "${meta.custid}" ? "${meta.custid}" : "${meta.id}" }
             publishDir = [
                 path: { "${params.outdir}/vcf2cytosure" },
                 mode: params.publish_dir_mode,

diff --git a/nextflow.config b/nextflow.config
@@ -74,6 +74,7 @@ params {
     reduced_penetrance              = null
     readcount_intervals             = null
     rtg_truthvcfs                   = null
+    sample_id_map                   = null
     sequence_dictionary             = null
     score_config_mt                 = null
     score_config_snv                = null

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -375,6 +375,17 @@
                     "help_text": "Should be Stranger's extended JSON as described at https://github.com/Clinical-Genomics/stranger/blob/master/stranger/resources/variant_catalog_grch37.json. This file is used by both ExpansionHunter and Stranger",
                     "fa_icon": "fas fa-file"
                 },
+                "sample_id_map": {
+                    "type": "string",
+                    "exists": true,
+                    "format": "file-path",
+                    "description": "Path to a file containing internal ids and customer ids in csv format.",
+                    "fa_icon": "fas fa-file-csv",
+                    "mimetype": "text/csv",
+                    "schema": "assets/sample_id_map.json",
+                    "help_text": "Optional file to rename sample ids in the vcf2cytosure vcf",
+                    "pattern": "^\\S+\\.csv$"
+                },
                 "vcf2cytosure_blacklist": {
                     "type": "string",
                     "help_text": "Optional file to blacklist regions for VCF2cytosure",

diff --git a/subworkflows/local/generate_cytosure_files.nf b/subworkflows/local/generate_cytosure_files.nf
@@ -2,36 +2,79 @@
 // Convert VCF with structural variations to the “.CGH” format used by the CytoSure Interpret Software
 //
 
-include { BCFTOOLS_VIEW as SPLIT_AND_FILTER_SV_VCF } from '../../modules/nf-core/bcftools/view/main'
-include { TIDDIT_COV as TIDDIT_COV_VCF2CYTOSURE    } from '../../modules/nf-core/tiddit/cov/main'
-include { VCF2CYTOSURE                             } from '../../modules/nf-core/vcf2cytosure/main'
+include { BCFTOOLS_VIEW as SPLIT_AND_FILTER_SV_VCF      } from '../../modules/nf-core/bcftools/view/main'
+include { BCFTOOLS_REHEADER as BCFTOOLS_REHEADER_SV_VCF } from '../../modules/nf-core/bcftools/reheader/main'
+include { TIDDIT_COV as TIDDIT_COV_VCF2CYTOSURE         } from '../../modules/nf-core/tiddit/cov/main'
+include { VCF2CYTOSURE                                  } from '../../modules/nf-core/vcf2cytosure/main'
 
 workflow GENERATE_CYTOSURE_FILES {
     take:
-        ch_vcf       // channel: [mandatory] [ val(meta), path(vcf) ]
-        ch_tbi       // channel: [mandatory] [ val(meta), path(vcf_index) ]
-        ch_bam       // channel: [mandatory] [ val(meta), path(bam) ]
-        ch_blacklist // channel: [optional] [path(blacklist)]
+        ch_vcf           // channel: [mandatory] [ val(meta), path(vcf) ]
+        ch_tbi           // channel: [mandatory] [ val(meta), path(vcf_index) ]
+        ch_bam           // channel: [mandatory] [ val(meta), path(bam) ]
+        ch_sample_id_map // channel: [optional] [val(id), val(id)]
+        ch_blacklist     // channel: [optional] [path(blacklist)]
 
     main:
-        ch_versions = Channel.empty()
+        ch_versions     = Channel.empty()
+        ch_reheader_out = Channel.empty()
 
         TIDDIT_COV_VCF2CYTOSURE (ch_bam, [[],[]])
 
         // Build channel: [val(sample_meta), path(vcf), path(vcf_index)]
         ch_vcf.join( ch_tbi, failOnMismatch: true )
             .set { ch_vcf_tbi }
 
-        ch_bam.combine(ch_vcf_tbi).map {
-            meta_sample, bam, meta_case, vcf, tbi ->
-            return [ meta_sample, vcf, tbi ]
-        }.set { ch_sample_vcf }
+        ch_bam.combine(ch_vcf_tbi)
+            .map {
+                meta_sample, bam, meta_case, vcf, tbi ->
+                new_meta = ['id':meta_sample.sample, 'sex':meta_sample.sex]
+                return [ new_meta, vcf, tbi ]
+            }
+            .join(ch_sample_id_map, remainder: true)
+            .branch { it  ->
+                id: it[3].equals(null)
+                    return [it[0] + [custid:it[0].id], it[1], it[2]]
+                custid: !(it[3].equals(null))
+                    return [it[0] + [custid:it[3]], it[1], it[2]]
+            }
+            .set { ch_for_mix }
+
+        Channel.empty()
+            .mix(ch_for_mix.id, ch_for_mix.custid)
+            .set { ch_sample_vcf }
 
         // Split vcf into sample vcf:s and frequency filter
         SPLIT_AND_FILTER_SV_VCF ( ch_sample_vcf, [], [], [] )
 
+        if (params.sample_id_map != null) {
+
+            SPLIT_AND_FILTER_SV_VCF.out.vcf
+                .map { meta, vcf -> return [meta, vcf, [], []]}
+                .set { ch_reheader_in }
+
+            BCFTOOLS_REHEADER_SV_VCF ( ch_reheader_in, [[:],[]] ).vcf
+                .set {ch_reheader_out}
+
+            ch_versions = ch_versions.mix(BCFTOOLS_REHEADER_SV_VCF.out.versions.first())
+        }
+
+        SPLIT_AND_FILTER_SV_VCF.out.vcf
+            .join(ch_reheader_out, remainder: true)
+            .branch { it  ->
+                split: it[2].equals(null)
+                    return [it[0], it[1]]
+                reheader: !(it[2].equals(null))
+                    return [it[0], it[2]]
+            }
+            .set { ch_for_mix }
+
+        Channel.empty()
+            .mix(ch_for_mix.split, ch_for_mix.reheader)
+            .set { ch_vcf2cytosure_in }
+
         VCF2CYTOSURE (
-            SPLIT_AND_FILTER_SV_VCF.out.vcf,
+            ch_vcf2cytosure_in,
             TIDDIT_COV_VCF2CYTOSURE.out.cov,
             [[:], []], [[:], []],
             ch_blacklist

diff --git a/workflows/raredisease.nf b/workflows/raredisease.nf
@@ -251,6 +251,8 @@ workflow RAREDISEASE {
                                                                             : Channel.value([])
     ch_rtg_truthvcfs            = params.rtg_truthvcfs                      ? Channel.fromPath(params.rtg_truthvcfs).collect()
                                                                             : Channel.value([])
+    ch_sample_id_map            = params.sample_id_map                      ? Channel.fromSamplesheet("sample_id_map")
+                                                                            : Channel.empty()
     ch_score_config_mt          = params.score_config_mt                    ? Channel.fromPath(params.score_config_mt).collect()
                                                                             : Channel.value([])
     ch_score_config_snv         = params.score_config_snv                   ? Channel.fromPath(params.score_config_snv).collect()
@@ -622,6 +624,7 @@ workflow RAREDISEASE {
             ch_sv_annotate.vcf_ann,
             ch_sv_annotate.tbi,
             ch_mapped.genome_marked_bam,
+            ch_sample_id_map,
             ch_vcf2cytosure_blacklist
         )
         ch_versions = ch_versions.mix(GENERATE_CYTOSURE_FILES.out.versions)