Add samples renaming for truth splitted files

nf-core · Nov 22, 2024 · cd450c3 · cd450c3
1 parent c1fd186
commit cd450c3
Show file tree

Hide file tree

Showing 5 changed files with 61 additions and 40 deletions.
diff --git a/conf/steps/validation.config b/conf/steps/validation.config
@@ -64,6 +64,20 @@ process {
         ext.args   = ["--ligate", "--output-type z", "--write-index=tbi"].join(' ')
     }
 
+    // Compute sample files for renaming
+    withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:BCFTOOLS_QUERY' {
+        tag = { "${meta.id} Batch ${meta.batch} ${meta.tools}" }
+        ext.args = '--list-samples'
+        publishDir = [enabled: false]
+    }
+
+    withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:GAWK' {
+        tag = { "${meta.id} Batch ${meta.batch} ${meta.tools}" }
+        ext.prefix = { "${meta.id}_samples"}
+        ext.args2 = "'BEGIN { OFS = \"\\t\" } { print \$1, \"-\", \$1\".truth\" }'"
+        publishDir = [enabled: false]
+    }
+
     // Split by samples
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:SPLIT_TRUTH:BCFTOOLS_PLUGINSPLIT' {
         ext.args   = ["--output-type z", "--write-index=tbi"].join(' ')

diff --git a/conf/test.config b/conf/test.config
@@ -13,7 +13,7 @@
 process {
     resourceLimits = [
         cpus: 4,
-        memory: '15.GB',
+        memory: '7.GB',
         time: '1.h'
     ]
 }

diff --git a/conf/test_all.config b/conf/test_all.config
@@ -13,7 +13,7 @@
 process {
     resourceLimits = [
         cpus: 2,
-        memory: '10.GB',
+        memory: '7.GB',
         time: '1.h'
     ]
 }

diff --git a/workflows/phaseimpute/main.nf b/workflows/phaseimpute/main.nf
@@ -65,6 +65,8 @@ include { BCFTOOLS_STATS as BCFTOOLS_STATS_TOOLS     } from '../../modules/nf-co
 
 // Concordance subworkflows
 include { BAM_GL_BCFTOOLS as GL_TRUTH                } from '../../subworkflows/local/bam_gl_bcftools'
+include { BCFTOOLS_QUERY                             } from '../../modules/nf-core/bcftools/query'
+include { GAWK                                       } from '../../modules/nf-core/gawk'
 include { VCF_SPLIT_BCFTOOLS as SPLIT_TRUTH          } from '../../subworkflows/local/vcf_split_bcftools'
 include { BCFTOOLS_STATS as BCFTOOLS_STATS_TRUTH     } from '../../modules/nf-core/bcftools/stats'
 include { VCF_CONCATENATE_BCFTOOLS as CONCAT_TRUTH   } from '../../subworkflows/local/vcf_concatenate_bcftools'
@@ -372,7 +374,7 @@ workflow PHASEIMPUTE {
         }
 
         // Split result by samples
-        VCF_SPLIT_BCFTOOLS(ch_input_validate)
+        VCF_SPLIT_BCFTOOLS(ch_input_validate.map{ [it[0], it[1], it[2], []] })
         ch_input_validate = VCF_SPLIT_BCFTOOLS.out.vcf_tbi
 
         // Compute stats on imputed files
@@ -443,8 +445,13 @@ workflow PHASEIMPUTE {
         CONCAT_TRUTH(ch_truth_vcf)
         ch_versions = ch_versions.mix(CONCAT_TRUTH.out.versions)
 
+        // Prepare renaming file
+        BCFTOOLS_QUERY(CONCAT_TRUTH.out.vcf_tbi, [], [], [])
+        GAWK(BCFTOOLS_QUERY.out.output, [])
+        ch_pluginsplit = CONCAT_TRUTH.out.vcf_tbi.join(GAWK.out.output.view())
+
         // Split truth vcf by samples
-        SPLIT_TRUTH(CONCAT_TRUTH.out.vcf_tbi)
+        SPLIT_TRUTH(ch_pluginsplit)
         ch_versions = ch_versions.mix(SPLIT_TRUTH.out.versions)
 
         // Compute stats on truth files

diff --git a/workflows/phaseimpute/tests/main.nf.test.snap b/workflows/phaseimpute/tests/main.nf.test.snap
@@ -25,10 +25,10 @@
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "24.04.4"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.1"
         },
-        "timestamp": "2024-10-25T16:47:19.203956394"
+        "timestamp": "2024-11-22T13:07:42.012993182"
     },
     "Check test_all": {
         "content": [
@@ -117,12 +117,12 @@
             [
                 "validation/concat/all.A-truth.vcf.gz",
                 "validation/concat/all.A-truth.vcf.gz.tbi",
-                "validation/samples/NA12878.vcf.gz",
-                "validation/samples/NA12878.vcf.gz.tbi",
-                "validation/samples/NA19401.vcf.gz",
-                "validation/samples/NA19401.vcf.gz.tbi",
-                "validation/samples/NA20359.vcf.gz",
-                "validation/samples/NA20359.vcf.gz.tbi",
+                "validation/samples/NA12878.truth.vcf.gz",
+                "validation/samples/NA12878.truth.vcf.gz.tbi",
+                "validation/samples/NA19401.truth.vcf.gz",
+                "validation/samples/NA19401.truth.vcf.gz.tbi",
+                "validation/samples/NA20359.truth.vcf.gz",
+                "validation/samples/NA20359.truth.vcf.gz.tbi",
                 "validation/stats/AllSamples.txt",
                 "validation/stats/NA12878.truth.bcftools_stats.txt",
                 "validation/stats/NA12878_P1000GP_Tglimpse1_SNP.txt",
@@ -167,19 +167,19 @@
         ],
         "meta": {
             "nf-test": "0.9.2",
-            "nextflow": "24.10.0"
+            "nextflow": "24.10.1"
         },
-        "timestamp": "2024-11-18T23:36:12.634065627"
+        "timestamp": "2024-11-22T13:58:54.188813202"
     },
     "Check test_validate": {
         "content": [
             [
-                "validation/samples/NA12878.vcf.gz",
-                "validation/samples/NA12878.vcf.gz.tbi",
-                "validation/samples/NA19401.vcf.gz",
-                "validation/samples/NA19401.vcf.gz.tbi",
-                "validation/samples/NA20359.vcf.gz",
-                "validation/samples/NA20359.vcf.gz.tbi",
+                "validation/samples/NA12878.truth.vcf.gz",
+                "validation/samples/NA12878.truth.vcf.gz.tbi",
+                "validation/samples/NA19401.truth.vcf.gz",
+                "validation/samples/NA19401.truth.vcf.gz.tbi",
+                "validation/samples/NA20359.truth.vcf.gz",
+                "validation/samples/NA20359.truth.vcf.gz.tbi",
                 "validation/stats/AllSamples.txt",
                 "validation/stats/NA12878.truth.bcftools_stats.txt",
                 "validation/stats/NA12878_SNP.txt",
@@ -191,9 +191,9 @@
         ],
         "meta": {
             "nf-test": "0.9.2",
-            "nextflow": "24.10.0"
+            "nextflow": "24.10.1"
         },
-        "timestamp": "2024-11-18T23:30:21.177496237"
+        "timestamp": "2024-11-22T13:54:09.435194577"
     },
     "Check test_batch": {
         "content": [
@@ -249,12 +249,12 @@
                 "NA20359"
             ],
             [
-                "validation/samples/NA12878.vcf.gz",
-                "validation/samples/NA12878.vcf.gz.tbi",
-                "validation/samples/NA19401.vcf.gz",
-                "validation/samples/NA19401.vcf.gz.tbi",
-                "validation/samples/NA20359.vcf.gz",
-                "validation/samples/NA20359.vcf.gz.tbi",
+                "validation/samples/NA12878.truth.vcf.gz",
+                "validation/samples/NA12878.truth.vcf.gz.tbi",
+                "validation/samples/NA19401.truth.vcf.gz",
+                "validation/samples/NA19401.truth.vcf.gz.tbi",
+                "validation/samples/NA20359.truth.vcf.gz",
+                "validation/samples/NA20359.truth.vcf.gz.tbi",
                 "validation/stats/AllSamples.txt",
                 "validation/stats/NA12878.truth.bcftools_stats.txt",
                 "validation/stats/NA12878_P1000GP_Tglimpse2_SNP.txt",
@@ -269,9 +269,9 @@
         ],
         "meta": {
             "nf-test": "0.9.2",
-            "nextflow": "24.10.0"
+            "nextflow": "24.10.1"
         },
-        "timestamp": "2024-11-18T23:47:48.301571495"
+        "timestamp": "2024-11-22T14:06:57.642618122"
     },
     "Check test_quilt": {
         "content": [
@@ -299,10 +299,10 @@
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "24.04.4"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.1"
         },
-        "timestamp": "2024-10-25T16:46:11.131198832"
+        "timestamp": "2024-11-22T13:05:58.709941089"
     },
     "Check test_sim": {
         "content": [
@@ -380,10 +380,10 @@
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "24.04.4"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.1"
         },
-        "timestamp": "2024-10-25T16:44:41.049965503"
+        "timestamp": "2024-11-22T13:03:28.516026252"
     },
     "Check test": {
         "content": [
@@ -463,9 +463,9 @@
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "24.04.4"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.1"
         },
-        "timestamp": "2024-10-25T16:56:33.72923425"
+        "timestamp": "2024-11-22T13:16:12.803136748"
     }
 }