optimized the resources and scratch dir usage across all the modules

Author: maurya-anand

modules/local/alignment/main.nf

@@ -25,7 +25,7 @@ process STAR_ALIGNMENT_WASP {
     --readFilesCommand zcat \\
     --outSAMtype BAM SortedByCoordinate \\
     --outSAMunmapped Within \\
-    --outTmpDir ${task.workDir} \\
+    --outTmpDir ${task.workDir}/_STARtmp \\
     --outFileNamePrefix ${meta.sampleid}. \\
     --outFilterMultimapNmax 20 \\
     --alignSJoverhangMin 8 \\
@@ -53,7 +53,7 @@ process STAR_ALIGNMENT_WASP {
     --chimMainSegmentMultNmax 1 \\
     --genomeLoad NoSharedMemory
 
-    samtools index ${meta.sampleid}.Aligned.sortedByCoord.out.bam
-    samtools stats ${meta.sampleid}.Aligned.sortedByCoord.out.bam > ${meta.sampleid}.samtools_stats.txt
+    samtools index -@ ${task.cpus} ${meta.sampleid}.Aligned.sortedByCoord.out.bam
+    samtools stats -@ ${task.cpus} ${meta.sampleid}.Aligned.sortedByCoord.out.bam > ${meta.sampleid}.samtools_stats.txt
     """
 }

modules/local/allele_count/main.nf

@@ -17,7 +17,7 @@ process ALLELE_COUNT{
 
     [[ -f "${reference_fa}.fai" ]] || samtools faidx ${reference_fa}
 
-    bgzip -c ${regions_vcf} > sites.vcf.gz
+    bgzip -@ ${task.cpus} -c ${regions_vcf} > sites.vcf.gz
 
     tabix -p vcf sites.vcf.gz
 
@@ -29,6 +29,7 @@ process ALLELE_COUNT{
         -R sites.vcf.gz \\
         -a FORMAT/AD,FORMAT/DP \\
         -Ou ${bam} \\
+        --threads ${task.cpus} \\
     | bcftools call -m -Ou -o ${meta.sampleid}.vcf
     bcftools query -f '%CHROM\\t%POS\\t%REF\\t%ALT\\t[%DP]\\t[%AD]\\n' ${meta.sampleid}.vcf \\
     | awk -F '\\t' 'BEGIN{OFS="\\t"}

modules/local/index_genome/main.nf

@@ -18,6 +18,6 @@ process STAR_GENOME_INDEX {
       --genomeDir STAR_INDEX \\
       --genomeFastaFiles ${ref_fa} \\
       --sjdbGTFfile ${gtf} \\
-      --outTmpDir ${task.workDir}
+      --outTmpDir ${task.workDir}/_STARgenomeTmp
     """
 }

modules/local/subset_1kgp/main.nf

@@ -13,7 +13,7 @@ process SUBSET_1KGP_VCF {
 
     total_threads=${task.cpus}
 
-    threads_per_job=\$(( total_threads > 4 ? 4 : total_threads ))
+    threads_per_job=\$(( total_threads > 2 ? 2 : total_threads ))
     parallel_jobs=\$(( total_threads / threads_per_job ))
     parallel_jobs=\$(( parallel_jobs < 1 ? 1 : parallel_jobs ))
 
@@ -55,7 +55,7 @@ process SUBSET_1KGP_VCF {
             [[ -s "${meta.sampleid}.chrX.snps.het.vcf.gz" ]] && echo "${meta.sampleid}.chrX.snps.het.vcf.gz"
         } > list.txt
 
-        bcftools concat -f list.txt -Ou \\
+        bcftools concat --threads \${total_threads} -f list.txt -Ou \\
             | bcftools sort --temp-dir ${task.workDir} -Ov -o ${meta.sampleid}.1KGP.snps.het.vcf
     else
         echo "Phased VCF dir is empty. Creating empty output files."

nextflow.config

@@ -21,28 +21,39 @@ profiles {
             errorStrategy = { task.exitStatus in [140, 143, 137, 104, 134, 139, 1] ? 'retry' : 'finish' }
             maxRetries = 3
             maxErrors = '-1'
-            cpus = { Math.min(4 * task.attempt, 64) }
-            memory = { "${Math.min(16 * task.attempt, 128)} GB" }
-            time = { "${Math.min(task.attempt, 3)}d" }
+            cpus = { Math.min(8 * task.attempt, 16) }
+            memory = { "${Math.min(12 * task.attempt, 24)} GB" }
+            time = { "${Math.min(2 * task.attempt, 6)}h" }
+            withName: 'CHECK_FASTQ' {
+                cpus = { Math.min(1 * task.attempt, 2) }
+                memory = { "${Math.min(2 * task.attempt, 4)} GB" }
+                time = { "${Math.min(1 * task.attempt, 2)}h" }
+            }
             withName: 'ADAPTER_TRIM' {
-                cpus = { Math.min(32 * task.attempt, 64) }
-                memory = { "${Math.min(64 * task.attempt, 128)} GB" }
-                time = { "${Math.min(task.attempt, 3)}d" }
+                cpus = { Math.min(8 * task.attempt, 32) }
+                memory = { "${Math.min(12 * task.attempt, 24)} GB" }
+                time = { "${Math.min(2 * task.attempt, 6)}h" }
             }
             withName: 'STAR_GENOME_INDEX' {
-                cpus = { Math.min(32 * task.attempt, 64) }
+                cpus = { Math.min(16 * task.attempt, 64) }
                 memory = { "${Math.min(48 * task.attempt, 128)} GB" }
-                time = { "${Math.min(task.attempt, 3)}d" }
+                time = { "${Math.min(2 * task.attempt, 4)}h" }
             }
             withName: 'STAR_ALIGNMENT_WASP' {
                 cpus = { Math.min(32 * task.attempt, 64) }
                 memory = { "${Math.min(64 * task.attempt, 128)} GB" }
-                time = { "${Math.min(task.attempt, 3)}d" }
+                time = { "${Math.min(3 * task.attempt, 6)}h" }
             }
             withName: 'SUBSET_1KGP_VCF' {
-                cpus = { Math.min(32 * task.attempt, 64) }
-                memory = { "${Math.min(32 * task.attempt, 128)} GB" }
-                time = { "${Math.min(task.attempt, 3)}d" }
+                stageInMode = 'symlink'
+                cpus = { Math.min(16 * task.attempt, 32) }
+                memory = { "${Math.min(64 * task.attempt, 128)} GB" }
+                time = { "${Math.min(2 * task.attempt, 6)}h" }
+            }
+            withName: 'REPORT' {
+                cpus = { Math.min(2 * task.attempt, 4) }
+                memory = { "${Math.min(4 * task.attempt, 8)} GB" }
+                time = { "${Math.min(30 * task.attempt, 120)}m" }
             }
         }
         workDir = "${System.getenv('WORK')}/nf-rna-wasp-allele-count_work"