fix genotyping connection to clonal analysis

Author: ggabernet

conf/test_genotyping.config

@@ -28,6 +28,6 @@ params {
     // Genotyping
     genotyping = true
     single_clone_representative = true
-    // Skip clonal analysis
-    skip_clonal_analysis = true
+    clonal_threshold = 0.1
+    skip_report_threshold = true
 }

subworkflows/local/clonal_analysis.nf

@@ -84,13 +84,16 @@ workflow CLONAL_ANALYSIS {
                                 it[0].species,
                                 it[0].single_cell,
                                 it[0].locus,
-                                it[1] ] }
+                                it[1],
+                                it[2] ] }
                 .groupTuple()
                 .map{ get_meta_tabs(it) }
-                .set{ ch_repertoire_groupped }
+                .set{ ch_repertoire_grouped }
+
+    ch_repertoire_grouped.dump(tag: "ch_repertoire_grouped")
 
     CLONAL_ASSIGNMENT(
-        ch_repertoire_reference,
+        ch_repertoire_grouped,
         clone_threshold.collect(),
         []
     )
@@ -104,6 +107,8 @@ workflow CLONAL_ANALYSIS {
             .map { it -> [ [id:'all_reps'], it ] }
             .set{ch_all_repertoires_cloned}
 
+    ch_all_repertoires_cloned.dump(tag: "ch_all_repertoires_cloned")
+
     if (!params.skip_all_clones_report){
 
         ch_all_repertoires_cloned_samplesheet = ch_all_repertoires_cloned.map{ it -> it[1] }
@@ -134,17 +139,23 @@ workflow CLONAL_ANALYSIS {
 
 // Function to map
 def get_meta_tabs(arr) {
+    if (arr[2].unique().size() > 1) {
+            error "Multiple subject_id found for ${arr[0]} (${arr[2].join(', ')}). Please check your input parameters and ensure that all samples with the same 'cloneby' value have the same 'subject_id' value."
+    }
+
     def meta = [:]
     meta.id            = [arr[0]].unique().join("")
     meta.sample_ids         = arr[1]
-    meta.subject_id         = arr[2]
-    meta.species            = arr[3]
+    meta.subject_id         = arr[2].unique().join("")
+    meta.species            = arr[3].unique().join("")
     meta.single_cell        = arr[4].unique().join("")
     meta.locus              = arr[5].unique().join("")
 
     def array = []
 
-        array = [ meta, arr[6].flatten() ]
-
+        array = [ meta, arr[6].flatten(), arr[7].unique() ]
+        if (arr[7].size() > 1) {
+            error "Multiple reference fasta files found for ${meta.id}. Please check your input parameters and ensure that all samples with the same ${params.genotypeby} value (parameter 'genotype_by') have the same ${params.cloneby} value (parameter 'clone_by')."
+        }
     return array
 }

subworkflows/local/novel_alleles_and_genotyping.nf

@@ -119,16 +119,24 @@ workflow NOVEL_ALLELES_AND_GENOTYPING {
 
 // Function to map
 def get_meta_tabs(arr) {
+    if (arr[2].unique().size() > 1) {
+        error "Multiple subject IDs found for ${arr[0]} (${arr[2].join(', ')}). It is not possible to perform joint genotyping of samples from different subjects. Please check the 'genotypeby' parameter."
+    }
+
     def meta = [:]
     meta.id            = [arr[0]].unique().join("")
     meta.sample_ids         = arr[1]
-    meta.subject_id         = arr[2]
-    meta.species            = arr[3]
+    meta.subject_id         = arr[2].unique().join("")
+    meta.species            = arr[3].unique().join("")
     meta.single_cell        = arr[4].unique().join("")
     meta.locus              = arr[5].unique().join("")
-
     def array = []
 
-    array = [ meta, arr[6].flatten(), arr[7][0] ]
+    array = [ meta, arr[6].flatten(), arr[7].unique() ]
+    if (arr[7].size() > 1) {
+        error "Multiple reference fasta files found for ${meta.id}."
+    }
+    
+
     return array
 }