Update module: sparsesignatures (#8950)

minor module updates

Co-authored-by: Simon Pearce <24893913+SPPearce@users.noreply.github.com>

Author: kdavydzenka

modules/nf-core/sparsesignatures/main.nf

@@ -9,7 +9,8 @@ process SPARSE_SIGNATURES {
 
     input:
     tuple val(meta), path(tsv_join,  stageAs: '*.tsv')
-
+    val(genome)   // genome version    
+ 
     output:
     tuple val(meta), path("*_mut_counts.rds"),            emit: signatures_mutCounts_rds
     tuple val(meta), path("*_cv_means_mse.rds"),          emit: signatures_cv_rds

modules/nf-core/sparsesignatures/meta.yml

@@ -38,6 +38,10 @@ input:
         pattern: "*.{tsv}"
         ontologies:
           - edam: http://edamontology.org/format_3475 # TSV
+  - genome:
+      type: string
+      description: Reference genome name to use with SparseSignatures (e.g. "GRCh37", "GRCh38")
+      ontologies: []
 output:
   signatures_mutCounts_rds:
     - - meta:
@@ -115,8 +119,8 @@ output:
         ontologies:
           - edam: http://edamontology.org/format_3750 # YAML
 authors:
-  - "@Katerina10-cloud"
+  - "@kdavydzenka"
   - "@elena-buscaroli"
 maintainers:
-  - "@Katerina10-cloud"
+  - "@kdavydzenka"
   - "@elena-buscaroli"

modules/nf-core/sparsesignatures/templates/main_script.R

@@ -21,7 +21,7 @@ parse_args <- function(x) {
 
 opt = list(
     prefix = ifelse('$task.ext.prefix' == 'null', '$meta.id', '$task.ext.prefix'),
-    genome = "NULL",
+    genome = "${genome}",
     K = "2:10",
     nmf_runs = "10",
     iterations = "30",
@@ -75,13 +75,19 @@ load_genome = function(genome, input_data) {
     if (genome == "GRCh37") {
         library(BSgenome.Hsapiens.1000genomes.hs37d5)
         bsg = BSgenome.Hsapiens.1000genomes.hs37d5::hs37d5
-        input_data <- input_data %>% mutate(chrom = str_remove(chrom,"^chr"))
+        
+	if (any(grepl("^chr", input_data[["chrom"]]))) {
+            input_data <- input_data %>% mutate(chrom = str_remove(chrom, "^chr"))
+        }
 
     } else if (genome == "GRCh38") {
         library(BSgenome.Hsapiens.UCSC.hg38)
         bsg = BSgenome.Hsapiens.UCSC.hg38
-
-        # Leave 'chrom' unchanged for GRCh38
+        
+	if (all(!grepl("^chr", input_data[["chrom"]]))) {
+            input_data <- input_data %>% mutate(chrom = paste0("chr", chrom))
+        }
+        
     }
     return(list(bsg = bsg, input_data = input_data))
 }

modules/nf-core/sparsesignatures/tests/main.nf.test

@@ -15,7 +15,7 @@ nextflow_process {
         when {
 
             params {
-              module_args = '[genome:GRCh37, K:2:3, nmf_runs:3, iterations:5, max_iterations_lasso:80, cross_validation_repetitions:3, cross_validation_iterations:3, cross_validation_entries:0.01, lambda_values_alpha:c(0.00, 0.01), lambda_values_beta:c(0.01, 0.05), lambda_rate_alpha:0, num_processes:all, seed:42]'
+              module_args = '[K:2:3, nmf_runs:3, iterations:5, max_iterations_lasso:80, cross_validation_repetitions:3, cross_validation_iterations:3, cross_validation_entries:0.01, lambda_values_alpha:c(0.00, 0.01), lambda_values_beta:c(0.01, 0.05), lambda_rate_alpha:0, num_processes:all, seed:42]'
             }
 
             process {
@@ -24,6 +24,7 @@ nextflow_process {
                     [ id:'test_grch37'],
                     file(params.modules_testdata_base_path + 'delete_me/sparsesignatures/test_GRCh37/MSeq_Set06_joint_table.tsv', checkIfExists: true)
                 ]
+                input[1] = "GRCh37" 
                 """
             }
         }
@@ -59,6 +60,7 @@ nextflow_process {
                     [ id:'test_grch37'], // meta map
                     file(params.modules_testdata_base_path + 'delete_me/sparsesignatures/test_GRCh37/MSeq_Set06_joint_table.tsv', checkIfExists: true)
                 ]
+                input[1] = "GRCh37" 
                 """
             }
         }
@@ -79,7 +81,7 @@ nextflow_process {
 
         when {
             params {
-                module_args = '[genome:GRCh38, K:2:3, nmf_runs:3, iterations:5, max_iterations_lasso:80, cross_validation_repetitions:3, cross_validation_iterations:3, cross_validation_entries:0.01, lambda_values_alpha:c(0.00, 0.01), lambda_values_beta:c(0.01, 0.05), lambda_rate_alpha:0, num_processes:all, seed:42]'
+                module_args = '[K:2:3, nmf_runs:3, iterations:5, max_iterations_lasso:80, cross_validation_repetitions:3, cross_validation_iterations:3, cross_validation_entries:0.01, lambda_values_alpha:c(0.00, 0.01), lambda_values_beta:c(0.01, 0.05), lambda_rate_alpha:0, num_processes:all, seed:42]'
             }
 
             process {
@@ -88,6 +90,7 @@ nextflow_process {
                     [ id:'test_grch38' ], // meta map
                     file(params.modules_testdata_base_path + 'delete_me/sparsesignatures/test_GRCh38/HCC1395_WXS_HCC1395_joint_table.tsv', checkIfExists: true)
                 ]
+                input[1] = "GRCh38" 
                 """
             }
         }
@@ -120,6 +123,7 @@ nextflow_process {
                     [ id:'test_grch38' ],
                     file(params.modules_testdata_base_path + 'delete_me/sparsesignatures/test_GRCh38/HCC1395_WXS_HCC1395_joint_table.tsv', checkIfExists: true)
                 ]
+                input[1] = "GRCh38" 
                 """
             }
         }