Refactor parameters validation

CHANGELOG.md

@@ -9,13 +9,14 @@ Initial release of genomic-medicine-sweden/skierfe, created with the [nf-core](h
 
 ### `Added`
 
+- Refactored conditionally required parameters validation [#69](https://github.com/genomic-medicine-sweden/skierfe/pull/69)
 - Added CNV workflow to test profile [#68](https://github.com/genomic-medicine-sweden/skierfe/pull/68)
-- Update and rename output directories [#65](https://github.com/genomic-medicine-sweden/skierfe/pull/65)
-- Update GLNexus version [#58](https://github.com/genomic-medicine-sweden/skierfe/pull/58)
+- Updated and rename output directories [#65](https://github.com/genomic-medicine-sweden/skierfe/pull/65)
+- Updated GLNexus version [#58](https://github.com/genomic-medicine-sweden/skierfe/pull/58)
 - Added uBAM support and multisample test [#51](https://github.com/genomic-medicine-sweden/skierfe/pull/51)
 - Added Revio BAM test data [#50](https://github.com/genomic-medicine-sweden/skierfe/pull/50)
-- Update template to 2.13.1 [#38](https://github.com/genomic-medicine-sweden/skierfe/pull/38)
-- Update pipeline to run with a small test dataset [#35](https://github.com/genomic-medicine-sweden/skierfe/pull/35)
+- Updated template to 2.13.1 [#38](https://github.com/genomic-medicine-sweden/skierfe/pull/38)
+- Updated pipeline to run with a small test dataset [#35](https://github.com/genomic-medicine-sweden/skierfe/pull/35)
 - Added test data and test profile [#33](https://github.com/genomic-medicine-sweden/skierfe/pull/33)
 
 ### `Fixed`

nextflow.config

@@ -15,7 +15,7 @@ params {
     dipcall_par                = null
     extra_snfs                 = null
     extra_gvcfs                = null
-    skip_qc                    = null
+    skip_qc                    = false
     skip_assembly_wf           = false
     skip_mapping_wf            = false
     skip_methylation_wf        = false

subworkflows/local/utils_nfcore_skierfe_pipeline/main.nf

@@ -20,6 +20,81 @@ include { imNotification            } from '../../nf-core/utils_nfcore_pipeline'
 include { UTILS_NFCORE_PIPELINE     } from '../../nf-core/utils_nfcore_pipeline'
 include { workflowCitation          } from '../../nf-core/utils_nfcore_pipeline'
 
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    DEFINE DEPENDENCIES (FILES AND WORKFLOWS) FOR OTHER WORKFLOWS
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+// nf-validation does not support contitional file and params validation,
+// add these here.
+//
+// For workflows:
+// - The cnv_calling workflow depends on the mapping and short_variant_calling workflows.
+//   It can't be run if either of --skip_mapping_wf or --skip_short_variant_calling is active.
+//
+// For files:
+// - Since the assembly workflow relies on the dipcall dipcall_par file,
+//   dipcall_par has to be set whenever --skip_assembly_wf is not active.
+//
+// For presets:
+// - Preset "pacbio" can't be run without --skip_methylation_wf.
+//
+
+def parameterDependencies = [
+    workflow: [
+        skip_assembly_wf          : [],
+        skip_short_variant_calling: ["skip_mapping_wf"],
+        skip_snv_annotation       : ["skip_mapping_wf", "skip_short_variant_calling"],
+        skip_cnv_calling          : ["skip_mapping_wf", "skip_short_variant_calling"],
+        skip_phasing_wf           : ["skip_mapping_wf", "skip_short_variant_calling"],
+        skip_repeat_wf            : ["skip_mapping_wf", "skip_short_variant_calling", "skip_phasing_wf"],
+        skip_methylation_wf       : ["skip_mapping_wf", "skip_short_variant_calling", "skip_phasing_wf"],
+    ],
+    files: [
+        dipcall_par    : ["skip_assembly_wf"],
+        snp_db         : ["skip_snv_annotation"],
+        vep_cache      : ["skip_snv_annotation"],
+        hificnv_xy     : ["skip_cnv_calling"],
+        hificnv_xx     : ["skip_cnv_calling"],
+        hificnv_exclude: ["skip_cnv_calling"],
+        trgt_repeats   : ["skip_repeat_wf"],
+    ],
+    preset: [
+        pacbio : ["skip_methylation_wf"],
+        ONT_R10: ["skip_assembly_wf", "skip_cnv_calling"],
+        revio  : [],
+    ]
+]
+
+def parameterStatus = [
+    workflow: [
+        skip_short_variant_calling: params.skip_short_variant_calling,
+        skip_phasing_wf           : params.skip_phasing_wf,
+        skip_methylation_wf       : params.skip_methylation_wf,
+        skip_repeat_wf            : params.skip_repeat_wf,
+        skip_snv_annotation       : params.skip_snv_annotation,
+        skip_cnv_calling          : params.skip_cnv_calling,
+        skip_mapping_wf           : params.skip_mapping_wf,
+        skip_qc                   : params.skip_qc,
+        skip_assembly_wf          : params.skip_assembly_wf,
+    ],
+    files: [
+        dipcall_par    : params.dipcall_par,
+        snp_db         : params.snp_db,
+        vep_cache      : params.vep_cache,
+        hificnv_xy     : params.hificnv_xy,
+        hificnv_xx     : params.hificnv_xx,
+        hificnv_exclude: params.hificnv_exclude,
+        trgt_repeats   : params.trgt_repeats,
+    ],
+    preset: [
+        pacbio : params.preset == "pacbio",
+        revio  : params.preset == "revio",
+        ONT_R10: params.preset == "ONT_R10",
+    ]
+]
+
 /*
 ========================================================================================
     SUBWORKFLOW TO INITIALISE PIPELINE
@@ -75,7 +150,7 @@ workflow PIPELINE_INITIALISATION {
     //
     // Custom validation for pipeline parameters
     //
-    validateInputParameters()
+    validateInputParameters(parameterDependencies, parameterStatus)
 
     //
     // Create channel from input file provided through params.input
@@ -137,8 +212,10 @@ workflow PIPELINE_COMPLETION {
 //
 // Check and validate pipeline parameters
 //
-def validateInputParameters() {
+
+def validateInputParameters(statusMap, paramsMap) {
     genomeExistsError()
+    validateParameterCombinations(statusMap, paramsMap)
 }
 
 //
@@ -228,3 +305,118 @@ def methodsDescriptionText(mqc_methods_yaml) {
 
     return description_html.toString()
 }
+
+//
+// Validate preset and workflow skip combinations
+//
+def validateParameterCombinations(combinationsMap, statusMap) {
+    // Array to store errors
+    def errors = []
+    // For each of the "workflow", "files", "preset"
+    statusMap.each { paramsType, allParams ->
+        // Go through all params and their status
+        statusMap[paramsType].each { param, paramStatus ->
+            switch (paramsType) {
+                case "files":
+                    checkFileDependencies(param, combinationsMap[paramsType], statusMap, errors)
+                case "workflow":
+                    checkWorkflowDependencies(param, combinationsMap[paramsType], statusMap, errors)
+                case "preset":
+                    checkPresetDependencies(param, combinationsMap[paramsType], statusMap, errors)
+                default:
+                    break
+            }
+        }
+    }
+    // Give error if there are any
+    if(errors) {
+        def error_string =
+            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n" +
+            "  " + errors.join("\n  ") + "\n" +
+            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"
+        error(error_string)
+    }
+}
+
+//
+// Lookup all other workflows that needs to be active for a certain workflow
+//
+def checkPresetDependencies(String preset, Map combinationsMap, Map statusMap, List errors) {
+    // Get the status of param
+    presetIsActive = statusMap["preset"][preset]
+    // If preset is not active, then give no error
+    if(!presetIsActive) {
+        return
+    }
+    // Get all required skips for a preset
+    def requiredSkips = combinationsMap[preset] as Set
+    // If there are no required skips, give no error
+    if(requiredSkips.size() == 0) {
+        return
+    }
+    // Collect the required --skips that are not active for the current preset
+    def dependencyString = findRequiredSkips("workflow", requiredSkips, statusMap)
+        .collect { "--$it" }
+        .join(" ")
+    errors << "--preset $preset is active, the pipeline has to be run with: $dependencyString"
+    return errors
+}
+
+def checkWorkflowDependencies(String workflow, Map combinationsMap, Map statusMap, List errors) {
+    // Get the status of param
+    paramStatus = statusMap["workflow"][workflow]
+    // If skip-parameter is not set, then workflow is active
+    activeWorkflow = !paramStatus
+    // If workflow is active, give no error
+    if(activeWorkflow) {
+        return
+    }
+    // Get all other worflows that need to be skipped when a certain workflow is skipped
+    def requiredWorkflows = combinationsMap.findAll { it.value.contains(workflow) }.keySet()
+    // If there are no required parameters, give no error
+    if(requiredWorkflows.size() == 0) {
+        return
+    }
+    // Collect the required --skips that are not active for the current workflow
+    def dependencyString = findRequiredSkips("workflow", requiredWorkflows, statusMap)
+        .collect { "--$it" }
+        .join(" ")
+
+    errors << "--$workflow is active, the pipeline has to be run with: $dependencyString"
+    return errors
+}
+//
+// Lookup if a file is required by any workflows, and add to errors
+//
+def checkFileDependencies(String fileParam, Map combinationsMap, Map statusMap, List errors) {
+    // Get the the workflow required by file (a file should only be required by one workflow)
+    def workflowThatRequiresFile = combinationsMap[fileParam][0]
+    // Get status of that workflow
+    def WorkflowIsOff = statusMap["workflow"][workflowThatRequiresFile]
+    def WorkflowIsActive = !WorkflowIsOff
+    // Get the file path
+    def FilePath = statusMap["files"][fileParam]
+    // If the workflow that requires the file is active & theres no file available
+    if(WorkflowIsActive && FilePath == null) {
+        errors << "--$workflowThatRequiresFile is NOT active, the following files are required: --$fileParam"
+    }
+    return errors
+}
+//
+// Find the workflow skips that are not active
+//
+def findRequiredSkips(presetType, Set<String> requiredWorkflows, Map statusMap) {
+    // Define a list to hold which additional parameters that need to be set
+    def skipsNeededToBeActive = []
+    // For all required workflows, check which are not skipped
+    for (requiredWorkflow in requiredWorkflows) {
+        // Get the status of the required workflow
+        requiredWorkflowIsSkipped = statusMap[presetType][requiredWorkflow]
+        // If that workflow is not skipped
+        if(!requiredWorkflowIsSkipped) {
+            // Add it to list of workflows that need to be skipped
+            skipsNeededToBeActive << requiredWorkflow
+        }
+    }
+    return skipsNeededToBeActive
+}