Patch/binsizes

.prettierignore

@@ -11,3 +11,4 @@ testing*
 *.pyc
 bin/
 .nf-test/
+null/

CHANGELOG.md

@@ -3,6 +3,13 @@
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## v1.1.1dev
+
+## Improvements
+
+1. Added several fail safe checks to the `--bin_sizes` parameter.
+2. Dynamically set the tag of the `WISECONDERX_NEWREF` depending on the `--bin_sizes` parameter.
+
 ## v1.1.0 - Naive Junior - [15 Sep 2023]
 
 ## New features

conf/modules.config

@@ -17,8 +17,9 @@ process {
     ]
 
     withName: WISECONDORX_NEWREF {
-        ext.prefix = {"${meta.id}_${meta.bin_size as Integer > 1000 ? meta.bin_size as Integer/1000 : meta.bin_size}${meta.bin_size as Integer > 1000 ? "k" : ""}bp"}
-        ext.args = { "--binsize ${meta.bin_size}" }
+        tag = { params.bin_sizes ? "${meta.bin_size}" : "${meta.id}" }
+        ext.prefix = { params.bin_sizes ? "${meta.id}_${meta.bin_size as Integer/1000}kbp" : "${meta.id}" }
+        ext.args = { params.bin_sizes ? "--binsize ${meta.bin_size}" : "" }
         publishDir = [
             enabled: true,
             path: { "${params.outdir}" },

conf/test.config

@@ -21,8 +21,6 @@ params {
 
     genomes_ignore = true
 
-    bin_sizes       = "10000,5"
-
     fasta           = params.test_data["homo_sapiens"]["genome"]["genome_fasta"]
     fai             = null //params.test_data["homo_sapiens"]["genome"]["genome_fasta_fai"]
 

nextflow.config

@@ -15,7 +15,7 @@ params {
 
     // Other options
     no_metrics                 = false
-    bin_sizes                  = "1000000,500000,50000,30000,15000,10000,5000,1000"
+    bin_sizes                  = null
 
     // References
     genome                     = "GRCh38"
@@ -234,7 +234,7 @@ manifest {
     description     = """A nextflow pipeline for creating references for WisecondorX"""
     mainScript      = 'main.nf'
     nextflowVersion = '!>=22.10.1'
-    version = '1.1.0'
+    version = '1.1.1dev'
     doi             = ''
 }
 

nextflow_schema.json

@@ -58,9 +58,16 @@
                     "description": "Don't create the metrics text file"
                 },
                 "bin_sizes": {
-                    "type": "string",
-                    "description": "A comma-delimited list of bin sizes to use for the analysis",
-                    "default": "1000000,500000,50000,30000,15000,10000,5000,1000"
+                    "anyOf": [
+                        {
+                            "type": "string",
+                            "description": "A comma-delimited list of bin sizes to use for the analysis"
+                        },
+                        {
+                            "type": "integer",
+                            "description": "A comma-delimited list of bin sizes to use for the analysis"
+                        }
+                    ]
                 }
             }
         },

tests/main.nf.test

@@ -4,19 +4,35 @@ nextflow_pipeline {
     script "main.nf"
     options "-stub"
 
-    test("Success") {
+    test("Success - No bin sizes") {
 
         when {
             params {
                 prefix = "test_reference"
-                bin_sizes = "10000,5"
+            }
+        }
+
+        then {
+            assert workflow.success
+            assert file("${outputDir}/test_reference.npz").exists()
+            assert file("${outputDir}/metrics.txt").exists()
+            assert file("${outputDir}/multiqc_reports/multiqc_report.html").exists()
+        }
+    }
+
+    test("Success - Bin sizes") {
+
+        when {
+            params {
+                prefix = "test_reference"
+                bin_sizes = "5000,10000"
             }
         }
 
         then {
             assert workflow.success
             assert file("${outputDir}/test_reference_10kbp.npz").exists()
-            assert file("${outputDir}/test_reference_5bp.npz").exists()
+            assert file("${outputDir}/test_reference_5kbp.npz").exists()
             assert file("${outputDir}/metrics.txt").exists()
             assert file("${outputDir}/multiqc_reports/multiqc_report.html").exists()
         }

workflows/cmgg-wisecondorx.nf

@@ -14,16 +14,21 @@ def summary_params = paramsSummaryMap(workflow)
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 */
 
-def checkPathParamList = [
-    params.input,
-    params.multiqc_config,
-    params.fasta,
-    params.fai
-]
-for (param in checkPathParamList) { if (param) { file(param, checkIfExists: true) } }
-
-// Check mandatory parameters
-if (params.input) { ch_input = file(params.input) } else { exit 1, 'Input samplesheet not specified!' }
+if(params.bin_sizes) {
+    val_bin_sizes = params.bin_sizes instanceof String ? params.bin_sizes.split(",") : [params.bin_sizes as String]
+    val_bin_sizes.each { bin_size ->
+        min_bin_size = 5000
+        divider = 5
+        if(bin_size.toInteger() < min_bin_size) {
+            error("Bin size ${bin_size} is too small, must be larger than or equal to ${min_bin_size}")
+        }
+        else if(bin_size.toInteger() % divider != 0){
+            error("Bin size ${bin_size} is not divisible by ${divider}")
+        }
+    }
+} else {
+    val_bin_sizes = [null]
+}
 
 /*
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -187,9 +192,9 @@ workflow CMGGWISECONDORX {
             [ new_meta, npz ]
         }
         .groupTuple() // All files should be present here, so no size is needed
-        .combine(params.bin_sizes.split(","))
+        .combine(val_bin_sizes)
         .map { meta, npz, bin_size ->
-            new_meta = meta + [bin_size:bin_size]
+            new_meta = bin_size ? meta + [bin_size:bin_size] : meta
             [ new_meta, npz ]
         }
         .set { ch_newref_input }