Merge fast processes

.nf-core.yml

@@ -43,4 +43,4 @@ template:
   skip_features:
     - fastqc
     - is_nfcore
-  version: 1.9.3
+  version: 1.10.0dev

conf/modules.config

@@ -49,15 +49,11 @@ process {
         ].join(" ")
     }
 
-    withName: "^.*CRAM_PREPARE_SAMTOOLS_BEDTOOLS:FILTER_BEDS\$" {
+    withName: "^.*CRAM_PREPARE_SAMTOOLS_BEDTOOLS:PROCESS_BEDS\$" {
         ext.prefix = { "${meta.id}.filter"}
         ext.args = "-vE \"LOW_COVERAGE|NO_COVERAGE${params.keep_alt_contigs ? "" : "|alt|random|decoy|Un"}\""
         ext.args2 = "-d 150"
-    }
-
-    withName: "^.*CRAM_PREPARE_SAMTOOLS_BEDTOOLS:BEDTOOLS_INTERSECT\$" {
-        ext.prefix = {"${meta.id}.intersect"}
-        ext.args = "-sorted"
+        ext.args3 = "-sorted"
     }
 
     /*
@@ -87,7 +83,7 @@ process {
 
     withName: "^.*CRAM_CALL_GATK4:VCF_CONCAT_BCFTOOLS:BCFTOOLS_CONCAT\$" {
         ext.prefix = { "${meta.id}.${meta.caller}.g" }
-        ext.args   = '--allow-overlaps --output-type z'
+        ext.args   = '--allow-overlaps --output-type z --write-index=tbi'
     }
 
     withName: "^.*CRAM_CALL_GATK4:BCFTOOLS_STATS\$" {
@@ -108,7 +104,7 @@ process {
 
     withName: "^.*BAM_CALL_ELPREP:VCF_CONCAT_BCFTOOLS:BCFTOOLS_CONCAT\$" {
         ext.prefix = { "${meta.id}.${meta.caller}.g" }
-        ext.args   = '--allow-overlaps --output-type z'
+        ext.args   = '--allow-overlaps --output-type z --write-index=tbi'
     }
 
     withName: "^.*BAM_CALL_ELPREP:BCFTOOLS_STATS\$" {
@@ -168,7 +164,7 @@ process {
 
     withName: "^.*GVCF_JOINT_GENOTYPE_GATK4:VCF_CONCAT_BCFTOOLS:BCFTOOLS_CONCAT\$" {
         ext.prefix = { "${meta.id}.concat" }
-        ext.args = "--allow-overlaps --output-type z"
+        ext.args = "--allow-overlaps --output-type z --write-index=tbi"
     }
 
     /*
@@ -194,14 +190,10 @@ process {
     }
 
     withName: "^.*BAM_CALL_VARDICTJAVA:VCF_CONCAT_BCFTOOLS:BCFTOOLS_CONCAT\$" {
-        ext.args   = '--allow-overlaps --output-type z'
+        ext.args   = '--allow-overlaps --output-type z --write-index=tbi'
         ext.prefix = {"${meta.id}.concat"}
     }
 
-    withName: "^.*BAM_CALL_VARDICTJAVA:TABIX_VCFANNO\$" {
-        ext.prefix = {"${meta.id}.vcfanno"}
-    }
-
     /*
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
         FILTER
@@ -210,28 +202,21 @@ process {
 
     withName: "^.*VCF_FILTER_BCFTOOLS:FILTER_1\$" {
         ext.prefix = { "${meta.id}.filtered1" }
-        ext.args = {
-            meta.caller == "vardict" ?
-                "-i 'QUAL >= 0${params.only_pass ? " && FILTER=\"PASS\"" : ""}' --output-type z":
-            meta.caller == "haplotypecaller" ?
-                "--output-type z --soft-filter 'GATKCutoffSNP' -e 'TYPE=\"snp\" && (MQRankSum < -12.5 || ReadPosRankSum < -8.0 || QD < 2.0 || FS > 60.0 || MQ < 30.0)' -m '+'":
-            meta.caller == "elprep" ?
-                "--output-type z --soft-filter 'GATKCutoffSNP' -e 'TYPE=\"snp\" && (MQRankSum < -12.5 || ReadPosRankSum < -8.0 || QD < 2.0 || FS > 60.0 || MQ < 30.0)' -m '+'":
-            ""
-        }
+        ext.args = { [
+            meta.caller == "vardict" ? "-i 'QUAL >= 0${params.only_pass ? " && FILTER=\"PASS\"" : ""}'" : "",
+            meta.caller == "haplotypecaller" ? "--soft-filter 'GATKCutoffSNP' -e 'TYPE=\"snp\" && (MQRankSum < -12.5 || ReadPosRankSum < -8.0 || QD < 2.0 || FS > 60.0 || MQ < 30.0)' -m '+'" : "",
+            meta.caller == "elprep" ? "--output-type z --soft-filter 'GATKCutoffSNP' -e 'TYPE=\"snp\" && (MQRankSum < -12.5 || ReadPosRankSum < -8.0 || QD < 2.0 || FS > 60.0 || MQ < 30.0)' -m '+'" : "",
+            "--output-type z --write-index=tbi"
+        ].findAll { arg -> arg != "" }.join(" ") }
     }
 
     withName: "^.*VCF_FILTER_BCFTOOLS:FILTER_2\$" {
-        ext.args = {
-            meta.caller == "vardict" ?
-                "--soft-filter 'LowFreqBias' --mode '+' -e  'FORMAT/AF[0:*] < 0.02 && FORMAT/VD[0] < 30 && INFO/SBF < 0.1 && INFO/NM >= 2.0' --output-type z" :
-            meta.caller == "haplotypecaller" ?
-                '--output-type z --soft-filter \'GATKCutoffIndel\' -e \'TYPE="indel" && (ReadPosRankSum < -20.0 || QD < 2.0 || FS > 200.0 || SOR > 10.0 )\' -m \'+\'' :
-            meta.caller == "elprep" ?
-                '--output-type z --soft-filter \'GATKCutoffIndel\' -e \'TYPE="indel" && (ReadPosRankSum < -20.0 || QD < 2.0 || FS > 200.0 || SOR > 10.0 )\' -m \'+\'' :
-            ""
-
-        }
+        ext.args = { [
+            meta.caller == "vardict" ? "--soft-filter 'LowFreqBias' --mode '+' -e  'FORMAT/AF[0:*] < 0.02 && FORMAT/VD[0] < 30 && INFO/SBF < 0.1 && INFO/NM >= 2.0'" : "",
+            meta.caller == "haplotypecaller" ? '--soft-filter \'GATKCutoffIndel\' -e \'TYPE="indel" && (ReadPosRankSum < -20.0 || QD < 2.0 || FS > 200.0 || SOR > 10.0 )\' -m \'+\'' : "",
+            meta.caller == "elprep" ? '--soft-filter \'GATKCutoffIndel\' -e \'TYPE="indel" && (ReadPosRankSum < -20.0 || QD < 2.0 || FS > 200.0 || SOR > 10.0 )\' -m \'+\'' : "",
+            "--output-type z --write-index=tbi"
+        ].findAll { arg -> arg != "" }.join(" ") }
         ext.prefix = {"${meta.id}.filtered"}
     }
 
@@ -243,7 +228,7 @@ process {
 
     withName: "^.*GERMLINE:BCFTOOLS_NORM\$" {
         ext.prefix = {"${meta.id}.normalized"}
-        ext.args = "-m-"
+        ext.args = "-m- --output-type z --write-index=tbi"
     }
 
     /*
@@ -264,7 +249,7 @@ process {
 
     withName: "^.*VCF_PED_RTGTOOLS:BCFTOOLS_ANNOTATE\$" {
         ext.prefix = { "${meta.id}.${meta.caller}.ped.annotated" }
-        ext.args = "--output-type z"
+        ext.args = "--output-type z --write-index=tbi"
     }
 
     /*
@@ -301,20 +286,16 @@ process {
         ].join(' ').trim()}
     }
 
-    withName: "^.*VCF_ANNOTATION:VCF_ANNOTATE_ENSEMBLVEP:BCFTOOLS_CONCAT\$" {
+    withName: "^.*VCF_ANNOTATE_ENSEMBLVEP:BCFTOOLS_CONCAT\$" {
         ext.prefix = {"${meta.id}_concat"}
         ext.args = "--allow-overlaps --output-type z"
     }
 
-    withName: "^.*VCF_ANNOTATION:VCF_ANNOTATE_ENSEMBLVEP:BCFTOOLS_SORT\$" {
+    withName: "^.*VCF_ANNOTATE_ENSEMBLVEP:BCFTOOLS_SORT\$" {
         ext.prefix = {"${meta.id}.sorted"}
+        ext.args = "--write-index=tbi --output-type z"
     }
 
-    withName: "^.*VCF_ANNOTATION:BGZIP_ANNOTATED_VCFS\$" {
-        ext.prefix = {"${meta.id}.vcfanno"}
-    }
-
-
     /*
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
         VALIDATION

modules.json

@@ -51,11 +51,6 @@
                         "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
                         "installed_by": ["modules"]
                     },
-                    "bedtools/intersect": {
-                        "branch": "master",
-                        "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
-                        "installed_by": ["modules"]
-                    },
                     "bedtools/merge": {
                         "branch": "master",
                         "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
@@ -191,16 +186,6 @@
                         "installed_by": ["modules"],
                         "patch": "modules/nf-core/somalier/relate/somalier-relate.diff"
                     },
-                    "tabix/bgzip": {
-                        "branch": "master",
-                        "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
-                        "installed_by": ["modules"]
-                    },
-                    "tabix/bgziptabix": {
-                        "branch": "master",
-                        "git_sha": "f448e846bdadd80fc8be31fbbc78d9f5b5131a45",
-                        "installed_by": ["modules"]
-                    },
                     "tabix/tabix": {
                         "branch": "master",
                         "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
@@ -225,7 +210,7 @@
                     },
                     "vcfanno": {
                         "branch": "master",
-                        "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
+                        "git_sha": "b1137e22798227331c9a9a12bd92bd6e865865c5",
                         "installed_by": ["modules"]
                     }
                 }

modules/local/filter_beds/main.nf → modules/local/process_beds/main.nf

@@ -1,4 +1,4 @@
-process FILTER_BEDS {
+process PROCESS_BEDS {
     tag "$meta.id"
     label 'process_single'
 
@@ -8,21 +8,27 @@ process FILTER_BEDS {
         'biocontainers/bedtools:2.31.0--hf5e1c6e_2' }"
 
     input:
-    tuple val(meta), path(bed)
+    tuple val(meta), path(bed), path(roi)
 
     output:
     tuple val(meta), path('*.bed'), emit: bed
     path  "versions.yml"          , emit: versions
 
     script:
-    // Remove regions with no coverage from the callable regions BED file
+    // Remove regions with no coverage from the callable regions BED file and intersect with an optional ROI file
     def args = task.ext.args ?: ''
     def args2 = task.ext.args2 ?: ''
+    def args3 = task.ext.args3 ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
 
     def unzip = bed.extension == "gz" ? "zcat" : "cat"
+    def intersect = roi ? "| bedtools intersect -a ${roi} -b - ${args3}" : ""
     """
-    ${unzip} ${bed} | grep ${args} | bedtools merge ${args2} > ${prefix}.bed
+    ${unzip} ${bed} \\
+        | grep ${args} \\
+        | bedtools merge ${args2} \\
+        ${intersect} \\
+        > ${prefix}.bed
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/local/filter_beds/tests/main.nf.test → modules/local/process_beds/tests/main.nf.test

@@ -1,12 +1,12 @@
 nextflow_process {
 
-    name "Test Process FILTER_BEDS"
+    name "Test Process PROCESS_BEDS"
     script "../main.nf"
-    process "FILTER_BEDS"
+    process "PROCESS_BEDS"
 
     tag "modules"
     tag "modules_local"
-    tag "filter_beds"
+    tag "PROCESS_BEDS"
 
     test("homo_sapiens - bed") {
 
@@ -17,7 +17,8 @@ nextflow_process {
                 """
                 input[0] = [
                     [ id:'test', single_end:false ], // meta map
-                    file(params.bed, checkIfExists:true)
+                    file(params.bed, checkIfExists:true),
+                    file(params.split1, checkIfExists:true)
                 ]
                 """
             }
@@ -42,7 +43,8 @@ nextflow_process {
                 """
                 input[0] = [
                     [ id:'test', single_end:false ], // meta map
-                    file(params.bed, checkIfExists:true)
+                    file(params.bed, checkIfExists:true),
+                    file(params.split1, checkIfExists:true)
                 ]
                 """
             }

modules/local/filter_beds/tests/main.nf.test.snap → modules/local/process_beds/tests/main.nf.test.snap

@@ -8,31 +8,31 @@
                             "id": "test",
                             "single_end": false
                         },
-                        "test.bed:md5,fa245abf8add7a80650566a1de67ec04"
+                        "test.bed:md5,ffd6f70b7fa85ffb46fb66a4cf9afb70"
                     ]
                 ],
                 "1": [
-                    "versions.yml:md5,38ea76fdc8d681b1e47415b195cccd88"
+                    "versions.yml:md5,c6be2bd45c01066eacc08f419bce67d6"
                 ],
                 "bed": [
                     [
                         {
                             "id": "test",
                             "single_end": false
                         },
-                        "test.bed:md5,fa245abf8add7a80650566a1de67ec04"
+                        "test.bed:md5,ffd6f70b7fa85ffb46fb66a4cf9afb70"
                     ]
                 ],
                 "versions": [
-                    "versions.yml:md5,38ea76fdc8d681b1e47415b195cccd88"
+                    "versions.yml:md5,c6be2bd45c01066eacc08f419bce67d6"
                 ]
             }
         ],
         "meta": {
             "nf-test": "0.9.1",
-            "nextflow": "24.10.0"
+            "nextflow": "24.10.3"
         },
-        "timestamp": "2024-11-13T17:12:06.470648263"
+        "timestamp": "2025-01-23T16:45:26.442970884"
     },
     "homo_sapiens - bed - stub": {
         "content": [
@@ -47,7 +47,7 @@
                     ]
                 ],
                 "1": [
-                    "versions.yml:md5,38ea76fdc8d681b1e47415b195cccd88"
+                    "versions.yml:md5,c6be2bd45c01066eacc08f419bce67d6"
                 ],
                 "bed": [
                     [
@@ -59,14 +59,14 @@
                     ]
                 ],
                 "versions": [
-                    "versions.yml:md5,38ea76fdc8d681b1e47415b195cccd88"
+                    "versions.yml:md5,c6be2bd45c01066eacc08f419bce67d6"
                 ]
             }
         ],
         "meta": {
             "nf-test": "0.9.1",
-            "nextflow": "24.10.0"
+            "nextflow": "24.10.3"
         },
-        "timestamp": "2024-11-13T17:09:43.531020092"
+        "timestamp": "2025-01-23T16:45:35.803097114"
     }
 }