Fix Nextflow I/O wiring: channels, lib imports, script execution

.gitignore

@@ -1,3 +1,8 @@
+# Nextflow runtime
+.nextflow/
+.nextflow.log*
+work/
+
 # Python cache
 __pycache__/
 *.pyc

main.nf

@@ -1,7 +1,5 @@
 #!/usr/bin/env nextflow
 
-nextflow.enable.dsl = 2
-
 include { PROMOTER_SEQUENCE_PIPELINE } from './workflows/main.nf'
 
 workflow {

modules/local/encode_and_split_pairs.nf

@@ -1,20 +1,23 @@
 process ENCODE_AND_SPLIT_PAIRS {
     tag { dataset_name }
 
-    publishDir { "${params.outdir}/${dataset_name}/Encoded" }, mode: params.publish_mode
+    publishDir "${params.outdir}/${dataset_name}/Encoded", mode: params.publish_mode
 
     input:
     tuple val(dataset_name), path(labeled_pairs)
     path human_fasta
     path mouse_fasta
     path original_gene_list
+    path project_lib
 
     output:
     tuple val(dataset_name), path("results/${dataset_name}/splits/${params.split_mode}"), emit: split_directory
 
     script:
+    def random_seqs_flag = params.random_seqs ? '--random-seqs' : ''
     """
-    python ${projectDir}/bin/encode_split.py \
+    export PYTHONPATH="${project_lib}/..:${project_lib}"
+    encode_split.py \
         --input ${labeled_pairs} \
         --name ${dataset_name} \
         --human-fasta ${human_fasta} \
@@ -24,6 +27,7 @@ process ENCODE_AND_SPLIT_PAIRS {
         --gene-list ${original_gene_list} \
         --val-frac ${params.val_frac} \
         --test-frac ${params.test_frac} \
-        --seed ${params.seed}
+        --seed ${params.seed} \
+        ${random_seqs_flag}
     """
 }

modules/local/label_gene_pairs.nf

@@ -1,17 +1,19 @@
 process LABEL_GENE_PAIRS {
     tag { dataset_name }
 
-    publishDir { "${params.outdir}/${dataset_name}/Labeling" }, mode: params.publish_mode
+    publishDir "${params.outdir}/${dataset_name}/Labeling", mode: params.publish_mode
 
     input:
     tuple val(dataset_name), path(raw_expression_dataset)
+    path project_lib
 
     output:
-    tuple val(dataset_name), path("results/${dataset_name}/Labeling/${params.profiling}_labeling_*"), emit: labeled_pairs
+    tuple val(dataset_name), path("results/${dataset_name}/Labeling/*.tsv"), emit: labeled_pairs
 
     script:
     """
-    python ${projectDir}/bin/label.py \
+    export PYTHONPATH="${project_lib}/..:${project_lib}"
+    label.py \
         --input ${raw_expression_dataset} \
         --name ${dataset_name} \
         --normalization ${params.normalization} \

modules/local/train_and_evaluate_model.nf

@@ -1,19 +1,20 @@
 process TRAIN_AND_EVALUATE_MODEL {
     tag "${dataset_name}"
 
-    publishDir {"${params.outdir}/${dataset_name}/Model"}, mode: params.publish_mode
+    publishDir "${params.outdir}/${dataset_name}/Model", mode: params.publish_mode
 
     input:
     tuple val(dataset_name), path(split_directory)
+    path project_lib
 
     output:
     tuple val(dataset_name), path("results/${dataset_name}/Model"), emit: model_directory
 
     script:
-    optimize_argument = params.optimize ? '--optimize' : ''
-
+    def optimize_argument = params.optimize ? '--optimize' : ''
     """
-    python ${projectDir}/bin/model_processing.py \
+    export PYTHONPATH="${project_lib}/..:${project_lib}"
+    model_processing.py \
         --data ${split_directory} \
         --name ${dataset_name} \
         --metric ${params.profiling} \

workflows/main.nf

@@ -1,43 +1,55 @@
-nextflow.enable.dsl = 2
-
 include { LABEL_GENE_PAIRS } from '../modules/local/label_gene_pairs.nf'
 include { ENCODE_AND_SPLIT_PAIRS } from '../modules/local/encode_and_split_pairs.nf'
 include { TRAIN_AND_EVALUATE_MODEL } from '../modules/local/train_and_evaluate_model.nf'
 
 workflow PROMOTER_SEQUENCE_PIPELINE {
 
+    // --- Validate required parameters ---
     if (params.input == null) {
         error "Missing required parameter: --input"
     }
-
     if (params.name == null) {
         error "Missing required parameter: --name"
     }
-
     if (params.human_fasta == null) {
         error "Missing required parameter: --human_fasta"
     }
-
     if (params.mouse_fasta == null) {
         error "Missing required parameter: --mouse_fasta"
     }
 
-    raw_expression_dataset_channel = Channel
+    // --- Stage the lib/ directory so Python scripts can import from it ---
+    ch_lib = Channel.value(file("${projectDir}/lib", checkIfExists: true))
+
+    // --- Input channels ---
+    // Queue channel: one tuple per dataset
+    ch_expression = Channel
         .fromPath(params.input, checkIfExists: true)
-        .map { raw_expression_dataset -> tuple(params.name, raw_expression_dataset) }
+        .map { f -> tuple(params.name, f) }
 
-    human_promoter_fasta_channel = Channel.fromPath(params.human_fasta, checkIfExists: true)
-    mouse_promoter_fasta_channel = Channel.fromPath(params.mouse_fasta, checkIfExists: true)
-    original_gene_list_channel = Channel.fromPath(params.input, checkIfExists: true)
+    // Value channels for single files (reusable across all items in the queue)
+    ch_human_fasta = Channel.value(file(params.human_fasta, checkIfExists: true))
+    ch_mouse_fasta = Channel.value(file(params.mouse_fasta, checkIfExists: true))
+    ch_gene_list   = Channel.value(file(params.input, checkIfExists: true))
 
-    LABEL_GENE_PAIRS(raw_expression_dataset_channel)
+    // --- Step 1: Label gene pairs ---
+    LABEL_GENE_PAIRS(
+        ch_expression,
+        ch_lib
+    )
 
+    // --- Step 2: Encode sequences and split into train/val/test ---
     ENCODE_AND_SPLIT_PAIRS(
         LABEL_GENE_PAIRS.out.labeled_pairs,
-        human_promoter_fasta_channel,
-        mouse_promoter_fasta_channel,
-        original_gene_list_channel
+        ch_human_fasta,
+        ch_mouse_fasta,
+        ch_gene_list,
+        ch_lib
     )
 
-    TRAIN_AND_EVALUATE_MODEL(ENCODE_AND_SPLIT_PAIRS.out.split_directory)
+    // --- Step 3: Train and evaluate the model ---
+    TRAIN_AND_EVALUATE_MODEL(
+        ENCODE_AND_SPLIT_PAIRS.out.split_directory,
+        ch_lib
+    )
 }