Merge pull request nf-core#86 from atrigila/bcftools_convert

Bcftools convert to vcf_sites_extract_bcftools
atrigila · Jun 11, 2024 · 150cb9e · 150cb9e
2 parents f7e5586 + 2d7f589
commit 150cb9e
Show file tree

Hide file tree

Showing 18 changed files with 221 additions and 136 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -32,6 +32,7 @@ Initial release of nf-core/phaseimpute, created with the [nf-core](https://nf-co
 - [#78](https://github.com/nf-core/phaseimpute/pull/78) - Separate validate step from panel preparation.
 - [#84](https://github.com/nf-core/phaseimpute/pull/84) - Change depth computation to use SAMTOOLS_DEPTH and make separation by chromosome only if regions are specified.
 - [#85](https://github.com/nf-core/phaseimpute/pull/85) - Use external params in individual tests for tools.
+- [#86](https://github.com/nf-core/phaseimpute/pull/86) - Move `bcftools_convert` to `vcf_sites_extract_bcftools`.
 
 ### `Fixed`
 

diff --git a/assets/schema_input_panel.json b/assets/schema_input_panel.json
@@ -28,16 +28,6 @@
                 "type": "string",
                 "pattern": "^\\S+\\.(vcf|bcf)(\\.gz)?\\.(tbi|csi)$",
                 "errorMessage": "Panel index file must be provided, cannot contain spaces and must have extension '.vcf' or '.bcf' with optional '.gz' extension and with '.csi' or '.tbi' extension"
-            },
-            "hap": {
-                "type": "string",
-                "pattern": "^\\S+\\.(hap)(\\.gz)?$",
-                "errorMessage": "Hap file can be provided, cannot contain spaces and must have extension '.hap' with '.gz' extension"
-            },
-            "legend": {
-                "type": "string",
-                "pattern": "^\\S+\\.(legend)(\\.gz)?$",
-                "errorMessage": "Legend file can be provided, cannot contain spaces and must have extension '.hap' with '.gz' extension"
             }
         },
         "required": ["panel", "chr", "vcf", "index"]

diff --git a/assets/schema_posfile.json b/assets/schema_posfile.json
@@ -33,6 +33,16 @@
                 "type": "string",
                 "pattern": "^\\S+\\.(txt|tsv)(\\.gz)?$",
                 "errorMessage": "TXT with sites (position file) per chromosome must be provided. Must have .txt or .tsv extension with optional .gz"
+            },
+            "hap": {
+                "type": "string",
+                "pattern": "^\\S+\\.(hap)(\\.gz)?$",
+                "errorMessage": "Hap file can be provided, cannot contain spaces and must have extension '.hap' with '.gz' extension"
+            },
+            "legend": {
+                "type": "string",
+                "pattern": "^\\S+\\.(legend)(\\.gz)?$",
+                "errorMessage": "Legend file can be provided, cannot contain spaces and must have extension '.hap' with '.gz' extension"
             }
         },
         "required": ["panel", "chr", "vcf", "index", "txt"]

diff --git a/conf/steps/imputation_glimpse1.config b/conf/steps/imputation_glimpse1.config
@@ -16,6 +16,7 @@ process {
     // Impute with GLIMPSE1
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_IMPUTE_GLIMPSE1:.*' {
         publishDir = [ enabled: false ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     // Call the variants before imputation

diff --git a/conf/steps/imputation_glimpse2.config b/conf/steps/imputation_glimpse2.config
@@ -14,6 +14,7 @@ process {
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_IMPUTE_GLIMPSE2:.*' {
         publishDir = [ enabled: false ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_IMPUTE_GLIMPSE2:GLIMPSE2_PHASE' {

diff --git a/conf/steps/imputation_quilt.config b/conf/steps/imputation_quilt.config
@@ -17,6 +17,7 @@ process {
             path: { "${params.outdir}/imputation/quilt/" },
             mode: params.publish_dir_mode,
         ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     // Impute quilt

diff --git a/conf/steps/imputation_stitch.config b/conf/steps/imputation_stitch.config
@@ -16,16 +16,17 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:POSFILE_PREPARE_GAWK:GUNZIP' {
-        ext.prefix = { "${meta.panel}_${meta.chr}_original_posfile" }
+        ext.prefix = { "${meta.id}_${meta.chr}_original_posfile" }
         publishDir = [enabled: false]
-        tag = {"${meta.panel}_${meta.chr}"}
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:POSFILE_PREPARE_GAWK:GAWK' {
         ext.args   = "'{ gsub(\",\", \"\\t\") ; key = \$1 FS \$2 } !seen[key]++'" // Remove duplicates
-        ext.prefix = { "${meta.panel}_${meta.chr}_posfile_stitch" }
+        ext.prefix = { "${meta.id}_${meta.chr}_posfile_stitch" }
         ext.suffix = "txt"
         publishDir = [enabled: false]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BAM_IMPUTE_STITCH:STITCH' {

diff --git a/conf/steps/panel_prep.config b/conf/steps/panel_prep.config
@@ -17,6 +17,7 @@ process {
             mode: params.publish_dir_mode,
             enabled: false
         ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_CHR_CHECK:VCF_CHR_RENAME:BCFTOOLS_ANNOTATE' {
@@ -35,6 +36,7 @@ process {
     // Subworkflow: VCF_NORMALIZE_BCFTOOLS
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_NORMALIZE_BCFTOOLS:.*' {
         publishDir = [ enabled: false ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_NORMALIZE_BCFTOOLS:BCFTOOLS_NORM' {
@@ -99,22 +101,14 @@ process {
         ]
     }
 
-    withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_NORMALIZE_BCFTOOLS:BCFTOOLS_CONVERT' {
-        ext.args = {"--haplegendsample ${meta.id}_${meta.chr}"}
-        publishDir = [
-            path: { "${params.outdir}/prep_panel/haplegend/" },
-            mode: params.publish_dir_mode,
-            enabled: true
-        ]
-    }
-
     // Subworkflow: VCF_PHASE_SHAPEIT5
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_PHASE_SHAPEIT5:.*' {
         publishDir = [
             path: { "${params.outdir}/prep_panel/phasing" },
             mode: params.publish_dir_mode,
             enabled: false
         ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_PHASE_SHAPEIT5:GLIMPSE2_CHUNK' {
@@ -126,8 +120,19 @@ process {
     }
 
     // Subworkflow: VCF_SITES_EXTRACT_BCFTOOLS
+
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_SITES_EXTRACT_BCFTOOLS:.*' {
         publishDir = [ enabled: false ]
+        tag = {"${meta.id} ${meta.chr}"}
+    }
+
+    withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_SITES_EXTRACT_BCFTOOLS:BCFTOOLS_CONVERT' {
+        ext.args = {"--haplegendsample ${meta.id}_${meta.chr}"}
+        publishDir = [
+            path: { "${params.outdir}/prep_panel/haplegend/" },
+            mode: params.publish_dir_mode,
+            enabled: true
+        ]
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_SITES_EXTRACT_BCFTOOLS:BCFTOOLS_VIEW' {
@@ -207,6 +212,7 @@ process {
             mode: params.publish_dir_mode,
             enabled: true
         ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_CHUNK_GLIMPSE:GLIMPSE_CHUNK' {

diff --git a/conf/steps/simulation.config b/conf/steps/simulation.config
@@ -13,6 +13,7 @@
 process {
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BAM_REGION:.*' {
         publishDir = [ enabled: false ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BAM_REGION:SAMTOOLS_VIEW' {
         ext.prefix = { "${meta.id}_R${meta.region.replace(':','_')}" }
@@ -23,10 +24,12 @@ process {
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
         ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BAM_DOWNSAMPLE:SAMTOOLS_DEPTH' {
         publishDir = [enabled: false]
-        ext.prefix = { "${meta.id}_C${meta.chr ?: "all"}.stats" }
+        ext.prefix = { "${meta1.id}_C${meta1.chr ?: "all"}.depth" }
+        tag = {"${meta1.id}"}
     }
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BAM_DOWNSAMPLE:SAMTOOLS_VIEW' {
         ext.prefix = { params.input_region ?
@@ -48,9 +51,11 @@ process {
     }
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BAM_DOWNSAMPLE:SAMTOOLS_MERGE' {
         ext.prefix = { "${meta.id}" }
+        tag = {"${meta.id}"}
     }
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BAM_DOWNSAMPLE:SAMTOOLS_INDEX_2' {
         ext.args = ""
+        tag = {"${meta.id}"}
     }
 
     // Coverage process

diff --git a/conf/steps/validation.config b/conf/steps/validation.config
@@ -16,6 +16,7 @@ process {
     // Compute genotype likelihoods for the truth set
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:GL_TRUTH:.*' {
         publishDir = [ enabled: false ]
+        tag = {"${meta.id} ${meta.chr}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:GL_TRUTH:BCFTOOLS_MPILEUP' {
@@ -65,6 +66,7 @@ process {
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
+        tag = {"${meta.id} ${meta.panel}"}
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_CONCORDANCE_GLIMPSE2:GLIMPSE2_CONCORDANCE' {
@@ -85,5 +87,6 @@ process {
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_CONCORDANCE_GLIMPSE2:GAWK' {
         ext.args   = "'(NR == 1) || (FNR > 1)'" // Skip header line
         ext.suffix = { "txt" }
+        tag = {"Test Quality"}
     }
 }
diff --git a/conf/test_quilt.config b/conf/test_quilt.config
@@ -32,7 +32,7 @@ params {
 
     // External params
     chunks = "${projectDir}/tests/csv/chunks.csv"
-    panel  = "${projectDir}/tests/csv/panel_2.csv"
+    posfile  = "${projectDir}/tests/csv/posfile.csv"
 
     // Impute tools
     tools  = "quilt"