311 add module gsea

conf/test_experimental.config

@@ -47,4 +47,9 @@ params {
 
     // Exploratory
     exploratory_main_variable   = 'contrasts'
+
+    // Activate GSEA
+    gsea_run = true
+    gene_sets_files = 'https://raw.githubusercontent.com/nf-core/test-datasets/modules/data/genomics/mus_musculus/gene_set_analysis/mh.all.v2022.1.Mm.symbols.gmt'
+
 }

subworkflows/local/enrichment/main.nf

@@ -4,13 +4,23 @@
 include { MYGENE } from "../../../modules/nf-core/mygene/main.nf"
 include { PROPR_GREA as GREA } from "../../../modules/local/propr/grea/main.nf"
 
+include { GSEA_GSEA } from '../../../modules/nf-core/gsea/gsea/main.nf'
+include { CUSTOM_TABULARTOGSEAGCT } from '../../../modules/nf-core/custom/tabulartogseagct/main.nf'
+include { CUSTOM_TABULARTOGSEACLS } from '../../../modules/nf-core/custom/tabulartogseacls/main.nf'
+include { TABULAR_TO_GSEA_CHIP } from '../../../modules/local/tabular_to_gsea_chip'
+
 workflow ENRICHMENT {
     take:
     ch_tools        // [ pathway_name, enrichment_map ]
     ch_counts
     ch_results_genewise
     ch_results_genewise_filtered
     ch_adjacency
+    ch_contrasts
+    ch_samplesheet
+    ch_featuresheet
+    ch_gene_sets
+    ch_versions
     // TODO: add ch_gm when provided by user, etc.
 
     main:
@@ -54,6 +64,58 @@ workflow ENRICHMENT {
 
     // todo: add gsea here
 
+    // For GSEA, we need to convert normalised counts to a GCT format for
+    // input, and process the sample sheet to generate class definitions
+    // (CLS) for the variable used in each contrast
+
+    CUSTOM_TABULARTOGSEAGCT ( ch_counts )
+
+    // TODO: update CUSTOM_TABULARTOGSEACLS for value channel input per new
+    // guidlines (rather than meta usage employed here)
+    ch_contrasts_and_samples = ch_contrasts
+        .map{it[0]} // revert back to contrasts meta map
+        .combine( ch_samplesheet.map { it[1] } )
+
+    CUSTOM_TABULARTOGSEACLS(ch_contrasts_and_samples)
+
+    TABULAR_TO_GSEA_CHIP(
+        ch_featuresheet.map{ it[1] },
+        [params.features_id_col, params.features_name_col]
+    )
+
+    // The normalised matrix does not always have a contrast meta, so we
+    // need a combine rather than a join here
+    // Also add file name to metamap for easy access from modules.config
+
+    ch_gsea_inputs = CUSTOM_TABULARTOGSEAGCT.out.gct
+        .map{ it.tail() }
+        .combine(CUSTOM_TABULARTOGSEACLS.out.cls)
+        .map{ tuple(it[1], it[0], it[2]) }
+        .combine(ch_gene_sets)
+
+    GSEA_GSEA(
+        ch_gsea_inputs,
+        ch_gsea_inputs.map{ tuple(it[0].reference, it[0].target) }, // *
+        TABULAR_TO_GSEA_CHIP.out.chip.first()
+    )
+
+    // * Note: GSEA module currently uses a value channel for the mandatory
+    // non-file arguments used to define contrasts, hence the indicated
+    // usage of map to perform that transformation. An active subject of
+    // debate
+    GSEA_GSEA.out.report_tsvs_ref.view()
+    ch_gsea_results = GSEA_GSEA.out.report_tsvs_ref
+        .join(GSEA_GSEA.out.report_tsvs_target)
+
+    ch_enriched = ch_enriched.combine(ch_gsea_results)
+
+
+    // Record GSEA versions
+    ch_versions = ch_versions
+        .mix(TABULAR_TO_GSEA_CHIP.out.versions)
+        .mix(GSEA_GSEA.out.versions)
+
+
     // ----------------------------------------------------
     // Perform enrichment analysis with gprofiler2
     // ----------------------------------------------------

subworkflows/local/experimental/main.nf

@@ -9,8 +9,11 @@ workflow EXPERIMENTAL {
     take:
     ch_contrasts    // [ meta, contrast_variable, reference, target ]
     ch_samplesheet  // [ meta, samplesheet ]
+    ch_featuresheet // [ meta, featuresheet ]
+    ch_gene_sets
     ch_counts       // [ meta, counts]
     ch_tools        // [ pathway_name, differential_map, correlation_map, enrichment_map ]
+    ch_versions
 
     main:
 
@@ -70,7 +73,12 @@ workflow EXPERIMENTAL {
         ch_counts,
         ch_results_genewise,
         ch_results_genewise_filtered,
-        ch_adjacency
+        ch_adjacency,
+        ch_contrasts,
+        ch_samplesheet,
+        ch_featuresheet,
+        ch_gene_sets,
+        ch_versions
     )
     ch_enriched = ch_enriched.mix(ENRICHMENT.out.enriched)
 

workflows/differentialabundance.nf

@@ -387,8 +387,11 @@ workflow DIFFERENTIALABUNDANCE {
         EXPERIMENTAL(
             ch_contrasts,
             VALIDATOR.out.sample_meta,
+            VALIDATOR.out.feature_meta,
+            ch_gene_sets,
             CUSTOM_MATRIXFILTER.out.filtered,
-            ch_tools
+            ch_tools,
+            ch_versions
         )
 
         // TODO for the moment, these channels are allocated to not breaking the next part.