Merge pull request #1439 from asp8200/replace_build_intervals_with_gawk

Replacing local module BUILD_INTERVALS with nf-core mdule GAWK
nf-core · Apr 3, 2024 · 52b3fff · 52b3fff
2 parents 2e06e65 + 2edcdb1
commit 52b3fff
Show file tree

Hide file tree

Showing 8 changed files with 118 additions and 32 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -29,6 +29,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - [#1426](https://github.com/nf-core/sarek/pull/1426) - Updating certain modules in order to fix the testdata-path in the nf-tests of those modules. Setting Docker runOptions for params.use_gatk_spark.
 - [#1428](https://github.com/nf-core/sarek/pull/1428) - Sync `TEMPLATE` with `tools` `2.13.1`
 - [#1431](https://github.com/nf-core/sarek/pull/1431) - Using docker.containerOptions instead of docker.runOptions. Clearing containerOptions for SPARK modules for any kind of supported container engine.
+- [#1439](https://github.com/nf-core/sarek/pull/1439) - Replacing the local module `BUILD_INTERVALS` with the nf-core module `GAWK`
 
 ### Fixed
 

diff --git a/conf/modules/prepare_intervals.config b/conf/modules/prepare_intervals.config
@@ -15,6 +15,11 @@
 
 process {
 
+    withName: 'BUILD_INTERVALS' {
+        ext.args   = { "-v FS='\t' -v OFS='\t' '{ print \$1, \"0\", \$2 }'" }
+        ext.suffix = { "bed" }
+    }
+
     withName: 'CREATE_INTERVALS_BED' {
         publishDir       = [
             mode: params.publish_dir_mode,

diff --git a/modules.json b/modules.json
@@ -294,6 +294,11 @@
                         "git_sha": "3f5420aa22e00bd030a2556dfdffc9e164ec0ec5",
                         "installed_by": ["modules"]
                     },
+                    "gawk": {
+                        "branch": "master",
+                        "git_sha": "dc3527855e7358c6d8400828754c0caa5f11698f",
+                        "installed_by": ["modules"]
+                    },
                     "manta/germline": {
                         "branch": "master",
                         "git_sha": "3f5420aa22e00bd030a2556dfdffc9e164ec0ec5",

diff --git a/modules/local/build_intervals/main.nf b/modules/local/build_intervals/main.nf
diff --git a/...les/local/build_intervals/environment.yml → modules/nf-core/gawk/environment.yml b/...les/local/build_intervals/environment.yml → modules/nf-core/gawk/environment.yml
diff --git a/modules/nf-core/gawk/main.nf b/modules/nf-core/gawk/main.nf
diff --git a/modules/nf-core/gawk/meta.yml b/modules/nf-core/gawk/meta.yml
diff --git a/subworkflows/local/prepare_intervals/main.nf b/subworkflows/local/prepare_intervals/main.nf
@@ -6,9 +6,9 @@
 // For all modules here:
 // A when clause condition is defined in the conf/modules.config to determine if the module should be run
 
-include { BUILD_INTERVALS                                        } from '../../../modules/local/build_intervals/main'
 include { CREATE_INTERVALS_BED                                   } from '../../../modules/local/create_intervals_bed/main'
 include { GATK4_INTERVALLISTTOBED                                } from '../../../modules/nf-core/gatk4/intervallisttobed/main'
+include { GAWK as BUILD_INTERVALS                                } from '../../../modules/nf-core/gawk/main'
 include { TABIX_BGZIPTABIX as TABIX_BGZIPTABIX_INTERVAL_SPLIT    } from '../../../modules/nf-core/tabix/bgziptabix/main'
 include { TABIX_BGZIPTABIX as TABIX_BGZIPTABIX_INTERVAL_COMBINED } from '../../../modules/nf-core/tabix/bgziptabix/main'
 
@@ -39,9 +39,9 @@ workflow PREPARE_INTERVALS {
     } else if (step != 'annotate' && step != 'controlfreec') {
         // If no interval/target file is provided, then generated intervals from FASTA file
         if (!intervals) {
-            BUILD_INTERVALS(fasta_fai.map{it -> [ [ id:it.baseName ], it ] })
+            BUILD_INTERVALS(fasta_fai.map{it -> [ [ id:it.baseName ], it ] }, [])
 
-            intervals_combined = BUILD_INTERVALS.out.bed
+            intervals_combined = BUILD_INTERVALS.out.output
 
             CREATE_INTERVALS_BED(intervals_combined.map{ meta, path -> path }, nucleotides_per_second)