refactor: use tagAlign for sicer and bam for macs2 & GEM

CCBR · Sep 11, 2023 · 2b24fb1 · 2b24fb1
1 parent 3467e83
commit 2b24fb1
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 15 deletions.
diff --git a/assets/dag.png b/assets/dag.png
diff --git a/main.nf b/main.nf
@@ -168,25 +168,37 @@ workflow {
 
   // peak calling
 
+  genome_frac = CALC_GENOME_FRAC(chrom_sizes)
+  // create channel with [ meta, chip_tag, input_tag, fraglen, genome_frac]
+  DEDUPLICATE.out.tag_align
+    .combine(DEDUPLICATE.out.tag_align)
+    .map {
+        meta1, tag1, meta2, tag2 ->
+            meta1.control == meta2.id ? [ meta1, tag1, tag2 ]: null
+    }
+    .join(frag_lengths)
+    .combine(genome_frac)
+    .set { ch_sicer }
+
   DEDUPLICATE.out.bam
     .combine(DEDUPLICATE.out.bam)
     .map {
         meta1, bam1, bai1, meta2, bam2, bai2 ->
             meta1.control == meta2.id ? [ meta1, bam1, bam2, [bai1, bai2] ]: null
     }
     .set { ch_ip_ctrl_bam }
-  // create channel with [ meta, chip_tag, input_tag, fraglen, genome_frac]
-  genome_frac = CALC_GENOME_FRAC(chrom_sizes)
   ch_ip_ctrl_bam
     .join(frag_lengths)
     .combine(genome_frac)
-    .set { ch_bam_macs }
-
-  ch_bam_macs | SICER
-  ch_bam_macs | MACS_BROAD
-  ch_bam_macs | MACS_NARROW
+    .set { ch_macs }
 
   ch_ip_ctrl_bam
     .combine(Channel.fromPath(params.gem_read_dists))
-    .combine(chrom_sizes) | GEM
+    .combine(chrom_sizes)
+    .set {ch_gem}
+
+  ch_sicer | SICER
+  ch_macs  | MACS_BROAD
+  ch_macs  | MACS_NARROW
+  ch_gem   | GEM
 }
diff --git a/modules/local/peaks.nf b/modules/local/peaks.nf
@@ -27,9 +27,10 @@ process SICER {
     label 'process_high'
 
     input:
-        tuple val(meta), path(chip), path(input), path(bais), val(fraglen), val(genome_frac)
+        tuple val(meta), path(chip), path(input), val(fraglen), val(genome_frac)
 
-    //output:
+    output:
+        tuple path("*.scoreisland"), path("*normalized.wig"), path("*islands-summary"), path("*island.bed")
 
     script:
     """
@@ -83,6 +84,7 @@ process MACS_BROAD {
 
     script:
     """
+    ls $chip > debug.txt
     macs2 callpeak \\
       -t ${chip} \\
       -c ${input} \\
@@ -142,7 +144,8 @@ process GEM {
     input:
         tuple val(meta), path(chip), path(input), path(bais), path(read_dists), path(chrom_sizes)
 
-    //output:
+    output:
+        path("*.GEM_events.narrowPeak")
 
     script:
     // $GEMJAR is defined in the docker container
@@ -158,7 +161,7 @@ process GEM {
       --k_max 13 \\
       --outNP \\
       --nrf \\
-      -f SAM
+      --f SAM
     """
 
     stub:

diff --git a/modules/local/qc.nf b/modules/local/qc.nf
@@ -187,7 +187,7 @@ process DEDUPLICATE {
         tuple val(meta), path(bam), path(chrom_sizes)
 
     output:
-        tuple val(meta), path("${meta.id}.TagAlign"), emit: tag_align
+        tuple val(meta), path("${meta.id}.TagAlign.bed"), emit: tag_align
         tuple val(meta), path("${bam.baseName}.dedup.bam"), path("${bam.baseName}.dedup.bam.bai"), emit: bam
         tuple path("${bam.baseName}.dedup.bam.flagstat"), path("${bam.baseName}.dedup.bam.idxstat"), emit: flagstat
 
@@ -196,8 +196,8 @@ process DEDUPLICATE {
     macs2 filterdup -i ${bam} -g ${params.align.effective_genome_size} --keep-dup="auto" -o TmpTagAlign1
     awk -F"\\t" -v OFS="\\t" '{{if (\$2>0 && \$3>0) {{print}}}}' TmpTagAlign1 > TmpTagAlign2
     awk -F"\\t" -v OFS="\\t" '{{print \$1,1,\$2}}' ${chrom_sizes} | sort -k1,1 -k2,2n > GenomeFile.bed
-    bedtools intersect -wa -f 1.0 -a TmpTagAlign2 -b GenomeFile.bed > ${meta.id}.TagAlign
-    bedtools bedtobam -i ${meta.id}.TagAlign -g ${chrom_sizes} | samtools sort -@ ${task.cpus} -o ${bam.baseName}.dedup.bam
+    bedtools intersect -wa -f 1.0 -a TmpTagAlign2 -b GenomeFile.bed > ${meta.id}.TagAlign.bed
+    bedtools bedtobam -i ${meta.id}.TagAlign.bed -g ${chrom_sizes} | samtools sort -@ ${task.cpus} -o ${bam.baseName}.dedup.bam
     samtools index ${bam.baseName}.dedup.bam
     samtools flagstat ${bam.baseName}.dedup.bam > ${bam.baseName}.dedup.bam.flagstat
     samtools idxstats ${bam.baseName}.dedup.bam > ${bam.baseName}.dedup.bam.idxstat