Merge branch 'aline'

Make deasy chaining smooth via bash instead to pure grrovy - add  use_slurm_for_aline to run script via sbatch for debugging purpose mainly

Author: Juke34

modules/aline.nf

@@ -7,6 +7,9 @@ process AliNe {
     tag "$pipeline_name"
     label 'aline'
     publishDir "${params.outdir}", mode: 'copy'
+    
+    errorStrategy 'terminate' // to avoid any retry
+    maxRetries 0  // Override global retry config - do not retry this process
 
     input:
         val pipeline_name     // String
@@ -27,13 +30,15 @@ process AliNe {
     script:
         def nxf_cmd = "nextflow run ${pipeline_name} ${profile} ${config} --reads ${reads} --reference ${genome} ${read_type} ${aligner} ${library_type} --annotation ${annotation} --data_type rna --outdir \$WORK_DIR/AliNe"
         """
+        echo "[AliNe] Process started at \$(date '+%Y-%m-%d %H:%M:%S')"
+
         # Save absolute work directory before changing context
         WORK_DIR=\$(pwd)
-
+        
         # Create cache directory for resume AliNe run made from different working directory
         mkdir -p "${cache_dir}"
         cd "${cache_dir}"
-
+        
         # Save command for reference/debugging
         echo "${nxf_cmd}" > \$WORK_DIR/nf-cmd.sh
 
@@ -47,38 +52,47 @@ process AliNe {
         #SBATCH --job-name=rain_AliNe_pipeline
         #SBATCH --cpus-per-task=1
         #SBATCH --mem=4G
+        #SBATCH --constraint=infiniband
         #SBATCH --time=2-00:00:00
         #SBATCH --output=\$WORK_DIR/aline_%j.out
         #SBATCH --error=\$WORK_DIR/aline_%j.err
 
         set -euo pipefail
 
-        cd "${cache_dir}"
-
+        # Load required modules
+        module load nextflow
+        module load singularity
+        
         echo "Starting AliNe pipeline at \$(date)"
         ${nxf_cmd}
         echo "AliNe pipeline completed at \$(date)"
         SBATCH_EOF
             
             # Submit job and capture job ID
             JOB_ID=\$(sbatch --parsable \$WORK_DIR/aline_job.sh)
-            echo "[AliNe] Submitted SLURM job: \$JOB_ID"
+            echo "[AliNe] Submitted SLURM job: \$JOB_ID at \$(date '+%Y-%m-%d %H:%M:%S')" 
             echo \$JOB_ID > \$WORK_DIR/aline_job_id.txt
             
             # Wait for job to appear in scheduler queue
-            echo "[AliNe] Waiting for job to appear in scheduler queue..."
-            sleep 5
-            
-            RETRY=0
-            while [ \$RETRY -lt 12 ]; do
+            # Simple wait for job to appear or to be started
+            echo "[AliNe] Waiting for job \$JOB_ID to appear in queue or to be started..."
+            while true; do
+                # Check if job is in queue
                 if squeue -j \$JOB_ID 2>/dev/null | grep -q \$JOB_ID; then
                     echo "[AliNe] Job \$JOB_ID is now visible in queue"
                     break
+                else
+                     echo "[AliNe] Job \$JOB_ID not yet visible in queue \$(date '+%Y-%m-%d %H:%M:%S')"
+                fi
+                # Check job state (crash test: if job has started or finished)
+                JOB_STATE=\$(sacct -j \$JOB_ID --format=State --noheader | head -1 | tr -d ' ')
+                if [[ "\$JOB_STATE" =~ ^(RUNNING|COMPLETED|FAILED|CANCELLED|TIMEOUT|PREEMPTED|NODE_FAIL|OUT_OF_MEMORY)\$ ]]; then
+                    echo "[AliNe] Job \$JOB_ID has state: \$JOB_STATE (not visible in queue, but started or finished)"
+                    break
                 fi
-                echo "[AliNe] Job not yet visible, waiting... (attempt \$((RETRY+1))/12)"
+                echo "[AliNe] Job not yet visible, waiting..."
                 sleep 5
-                RETRY=\$((RETRY+1))
-            done
+            done            # Simple wait for job to appear or to be started
             
             # Wait for job completion
             echo "[AliNe] Waiting for job \$JOB_ID to complete..."
@@ -88,35 +102,36 @@ process AliNe {
             
             # Check job exit status
             JOB_STATE=\$(sacct -j \$JOB_ID --format=State --noheader | head -1 | tr -d ' ')
-            echo "[AliNe] Job \$JOB_ID finished with state: \$JOB_STATE"
+            echo "[AliNe] Job \$JOB_ID finished with state: \$JOB_STATE at \$(date '+%Y-%m-%d %H:%M:%S')"
             
             if [[ "\$JOB_STATE" != "COMPLETED" ]]; then
-                echo "[AliNe] ERROR: Job failed with state \$JOB_STATE" >&2
-                cat \$WORK_DIR/aline_*.err >&2 || true
+                echo "[AliNe] ERROR: Job failed with state \$JOB_STATE at \$(date '+%Y-%m-%d %H:%M:%S')" >&2
+                echo "With message (100 last lines)": >&2
+                tail -n 100  \$WORK_DIR/aline_\$JOB_ID.out >&2
                 exit 1
             fi
             
             # Copy log for reference
             if [ -f .nextflow.log ]; then
                 cp .nextflow.log \$WORK_DIR/nextflow.log
             fi
-            
-            echo "[AliNe] Pipeline completed successfully via SLURM"
+
+            echo "[AliNe] Pipeline completed successfully via SLURM at \$(date '+%Y-%m-%d %H:%M:%S')"
         else
             echo "[AliNe] Detected local/standard environment - running AliNe directly"
             
             # Run nextflow command directly
-            ${nxf_cmd}
-            
-            # Copy log for reference
-            if [ -f .nextflow.log ]; then
-                cp .nextflow.log \$WORK_DIR/nextflow.log
-            fi
+            ${nxf_cmd} || {
+                echo "[AliNe] ERROR: Pipeline failed at \$(date '+%Y-%m-%d %H:%M:%S')" >&2
+                exit 1
+            }
             
-            echo "[AliNe] Pipeline completed successfully (direct execution)"
+            echo "[AliNe] Pipeline completed successfully (direct execution) at \$(date '+%Y-%m-%d %H:%M:%S')"
         fi
+        
+        echo "[AliNe] Process finished at \$(date '+%Y-%m-%d %H:%M:%S')"
 """
 
     output:
         path "AliNe"  , emit: output
-}
+}

rain.nf

@@ -35,6 +35,7 @@ params.region = "" // e.g. chr21 - Used to limit the analysis to a specific regi
 params.help = null
 params.monochrome_logs = false // if true, no color in logs
 params.debug = false // Enable debug output
+params.use_slurm_for_aline = false // Whether to submit AliNe as a separate SLURM job when using an HPC environment
 
 // --------------------------------------------------
 /* ---- Params shared between RAIN and AliNe ---- */
@@ -61,7 +62,7 @@ params.trimming_fastp = false
 align_tools = [ 'bbmap', 'bowtie', 'bowtie2', 'bwaaln', 'bwamem', 'bwamem2', 'bwasw', 'dragmap', 'graphmap2', 'hisat2', 'kallisto', 'last', 'minimap2', 'novoalign', 'nucmer', 'ngmlr', 'salmon', 'star', 'subread', 'sublong' ]
 params.aligner = 'hisat2'
 // AliNe version
-params.aline_version = 'v1.6.1'
+params.aline_version = 'v1.6.2'
 //*************************************************
 // STEP 1 - HELP
 //*************************************************
@@ -206,15 +207,11 @@ else { exit 1, "No executer selected: please use a profile activating docker or
 
 // check AliNE profile
 def aline_profile_list=[]
-def use_slurm_for_aline = false
+def use_slurm_for_aline = params.use_slurm_for_aline
 str_list = workflow.profile.tokenize(',')
 str_list.each {
     if ( it in aline_profile_allowed ){
          aline_profile_list.add(it)
-         if (it == 'itrop') {
-             // Don't pass itrop to AliNe, but remember we're in slurm context
-             use_slurm_for_aline = true
-         } 
     }
 }
 def aline_profile = aline_profile_list.join(',')