fix: use collectFile to avoid filename collisions in prep_diffbind

main.nf

@@ -121,11 +121,12 @@ workflow CHIPSEQ {
         if (params.contrasts) {
             contrasts = file(params.contrasts, checkIfExists: true)
             // TODO use consensus peaks for regions of interest in diffbind
-            bam_peaks = CALL_PEAKS.out.bam_peaks
+            CALL_PEAKS.out.bam_peaks
                 .combine(deduped_bam)
                 .map{meta1, bam1, bai1, peak, tool, meta2, bam2, bai2 ->
                     meta1.control == meta2.id ? [ meta1.sample_basename, meta1 + [tool: tool], bam1, bai1, peak, bam2, bai2 ] : null
                 }
+                .set{bam_peaks}
             DIFF( bam_peaks, INPUT_CHECK.out.csv, contrasts )
 
         }

modules/local/check_contrasts/templates/check_contrasts.R

@@ -23,13 +23,15 @@ main <- function(contrasts_filename = "${contrasts}",
   )
 }
 
-#' Ensure contrast has exactly two groups to compare
+#' Check an individual contrast comparison
 check_contrast <- function(contrast_name, contrasts_lst) {
+  # Ensure contrast has exactly two groups to compare
   contrast_len <- length(contrasts_lst[[contrast_name]])
   assert_that(
     contrast_len == 2,
     msg = glue("Contrasts must have only two groups per comparison, but {contrast_name} has {contrast_len}")
   )
+  # TODO: check that each sample only appears in one group
 }
 
 #' Combine sample sheet with contrast group

modules/local/diffbind/prep/main.nf

@@ -1,30 +1,26 @@
 process PREP_DIFFBIND {
-    tag { "${metas[0].contrast}.${metas[0].tool}" }
+    tag { "${meta.id}.${meta.contrast}.${meta.tool}" }
     label 'process_single'
 
+    container "${params.containers.base}"
+
     input:
-        tuple val(metas), path(bams), path(bais), path(peaks), path(ctrl_bams), path(ctrl_bais)
+        tuple val(meta), path(bam), path(bai), path(peak), path(ctrl_bam), path(ctrl_bai)
 
     output:
-        tuple val(new_meta), path("*.csv"), emit: csv
+        tuple val(meta), path("*.csv"), emit: csv
 
     script:
-    assert metas*.contrast.toSet().size() == 1
-    assert metas*.tool.toSet().size() == 1
-    new_meta = [contrast: metas[0].contrast, tool: metas[0].tool]
-    def csv_text = [['SampleID', "Replicate", 'Condition', 'bamReads', "ControlID", "bamControl", 'Peaks', 'PeakCaller']]
-    [metas, bams, bais, peaks]
-        .transpose()
-        .each { meta, bam, bai, peak ->
-            csv_text.add([meta.id, meta.rep, meta.group, bam, meta.control, ctrlbam, peak, meta.tool])
-        }
-    csv_text = csv_text*.join(',').join(System.lineSeparator)
+    def csv_text = [
+        ['SampleID', "Replicate", 'Condition', 'bamReads', "ControlID", "bamControl", 'Peaks', 'PeakCaller'],
+        [meta.id,     meta.rep,    meta.group,  bam,       meta.control, ctrl_bam,     peak,    meta.tool]
+    ]*.join(',').join(System.lineSeparator)
     """
-    echo -ne "${csv_text}" > ${metas[0].contrast}.${metas[0].tool}.csv
+    echo -ne "${csv_text}" > ${meta.contrast}.${meta.tool}.${meta.id}.csv
     """
 
     stub:
     """
-    touch ${metas[0].contrast}.${metas[0].tool}.csv
+    touch ${meta.contrast}.${meta.tool}.${meta.id}.csv
     """
 }

subworkflows/local/diff/main.nf

@@ -22,16 +22,15 @@ workflow DIFF {
         bam_peaks
             .combine( contrasts )
             .map{ sample_basename1, peak_meta, bam, bai, peak, ctrl_bam, ctrl_bai, sample_basename2, con_meta ->
-                sample_basename1 == sample_basename2 ? [ "${meta.contrast}.${meta.tool}", peak_meta + con_meta, bam, bai, peak, ctrl_bam, ctrl_bai ] : null
+                sample_basename1 == sample_basename2 ? [ peak_meta + con_meta, bam, bai, peak, ctrl_bam, ctrl_bai ] : null
             }
             .unique()
-            .groupTuple()
-            .map{ it -> // drop meta.contrast
-              it[1..-1]
-            }
             .set{ ch_peaks_contrasts }
-        ch_peaks_contrasts | view
         ch_peaks_contrasts | PREP_DIFFBIND
+        PREP_DIFFBIND.out.csv
+            .collectFile(storeDir: "${params.outdir}/diffbind/contrasts") { meta, row ->
+                [ "${meta.contrast}.${meta.tool}.csv", row + '\n' ]
+            }
     emit:
         diff_peaks = bam_peaks