stjude · madetunj · May 18, 2023 · May 18, 2023 · May 18, 2023
diff --git a/peaseq-case.wdl b/peaseq-case.wdl
@@ -383,10 +383,16 @@ workflow peaseq {
             #   Summary statistics on FASTQs
             #   Combine html files into one for easy viewing
 
+            call util.basicfastqstats as indv_R1_bfs {
+                input :
+                    fastqfile=fastqpair.left,
+                    default_location='SAMPLE/' + sub(basename(fastqpair.left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats'
+            }
             call mapping.mapping as indv_PE_mapping {
                 input :
                     fastqfile=fastqpair.left,
                     fastqfile_R2=fastqpair.right,
+                    metricsfile=indv_R1_bfs.metrics_out,
                     insert_size=insertsize,
                     strandedness=strandedness,
                     index_files=actual_bowtie_index,
@@ -526,10 +532,16 @@ workflow peaseq {
         #   Summary statistics on FASTQs
         #   Combine html files into one for easy viewing
 
+        call util.basicfastqstats as R1_bfs {
+            input :
+                fastqfile=sample_fastqfiles[0].left,
+                default_location=if defined(results_name) then results_name + '/BAM_files' else sub(basename(sample_fastqfiles[0].left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats'
+        }
         call mapping.mapping as uno_PE_mapping {
             input :
                 fastqfile=sample_fastqfiles[0].left,
                 fastqfile_R2=sample_fastqfiles[0].right,
+                metricsfile=R1_bfs.metrics_out,
                 insert_size=insertsize,
                 strandedness=strandedness,
                 index_files=actual_bowtie_index,

diff --git a/peaseq-control.wdl b/peaseq-control.wdl
@@ -567,10 +567,17 @@ workflow peaseq {
             #   Summary statistics on FASTQs
             #   Combine html files into one for easy viewing
 
+            call util.basicfastqstats as s_indv_R1_bfs {
+                input :
+                    fastqfile=fastqpair.left,
+                    default_location=if multi_fastqpair then 'SAMPLE/individual_fastqs/' + sub(basename(fastqpair.left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats' else 'SAMPLE/' + sub(basename(fastqpair.left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats'
+            }
+
             call mapping.mapping as s_indv_PE_mapping {
                 input :
                     fastqfile=fastqpair.left,
                     fastqfile_R2=fastqpair.right,
+                    metricsfile=s_indv_R1_bfs.metrics_out,
                     insert_size=insertsize,
                     strandedness=strandedness,
                     index_files=actual_bowtie_index,
@@ -702,10 +709,17 @@ workflow peaseq {
             #   Summary statistics on FASTQs
             #   Combine html files into one for easy viewing
 
+            call util.basicfastqstats as c_indv_R1_bfs {
+                input :
+                    fastqfile=fastqpair.left,
+                    default_location=if multi_fastqpair then 'CONTROL/individual_fastqs/' + sub(basename(fastqpair.left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats' else 'CONTROL/' + sub(basename(fastqpair.left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats'
+            }
+
             call mapping.mapping as c_indv_PE_mapping {
                 input :
                     fastqfile=fastqpair.left,
                     fastqfile_R2=fastqpair.right,
+                    metricsfile=c_indv_R1_bfs.metrics_out,
                     insert_size=insertsize,
                     strandedness=strandedness,
                     index_files=actual_bowtie_index,
@@ -845,10 +859,18 @@ workflow peaseq {
         #   Summary statistics on FASTQs
         #   Combine html files into one for easy viewing
 
+        call util.basicfastqstats as s_R1_bfs {
+            input :
+                fastqfile=sample_fastqfiles[0].left,
+                default_location=if multi_fastqpair then 'SAMPLE/individual_fastqs/' + sub(basename(sample_fastqfiles[0].left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats' else 'SAMPLE/' + sub(basename(sample_fastqfiles[0].left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats'
+
+        }
+
         call mapping.mapping as s_uno_PE_mapping {
             input :
                 fastqfile=sample_fastqfiles[0].left,
                 fastqfile_R2=sample_fastqfiles[0].right,
+                metricsfile=s_R1_bfs.metrics_out,
                 insert_size=insertsize,
                 strandedness=strandedness,
                 index_files=actual_bowtie_index,
@@ -883,10 +905,17 @@ workflow peaseq {
         #   Summary statistics on FASTQs
         #   Combine html files into one for easy viewing
 
+        call util.basicfastqstats as c_R1_bfs {
+            input :
+                fastqfile=control_fastqfiles[0].left,
+                default_location=if multi_control_fastqpair then 'CONTROL/individual_fastqs/' + sub(basename(control_fastqfiles[0].left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats' else 'CONTROL/' + sub(basename(control_fastqfiles[0].left),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','') + '/QC/SummaryStats'
+        }
+
         call mapping.mapping as c_uno_PE_mapping {
             input :
                 fastqfile=control_fastqfiles[0].left,
                 fastqfile_R2=control_fastqfiles[0].right,
+                metricsfile=c_R1_bfs.metrics_out,
                 insert_size=insertsize,
                 strandedness=strandedness,
                 index_files=actual_bowtie_index,

diff --git a/workflows/tasks/bowtie.wdl b/workflows/tasks/bowtie.wdl
@@ -8,7 +8,6 @@ task bowtie {
         Array[File]+ index_files
         String? prefix
         String outputfile = if (defined(prefix)) then select_first([prefix,basename(fastqfile)]) + '.sam' else if (defined(fastqfile_R2)) then sub(basename(fastqfile),'_R?[12]_....f.*q.gz|_R?[12].f.*q.gz','.sam') else sub(basename(fastqfile),'.fastq.gz|.fq.gz','.sam')
-
         Int read_length = 75
         Int insert_size = 600
         Int limit_alignments = 2
@@ -29,15 +28,18 @@ task bowtie {
 
     command <<<
         if [ -f "~{metricsfile}" ]; then
-            readlength=$(tail -n 1 ~{metricsfile} | awk '{print $4}');
+            readlength=$(tail -n 1 ~{metricsfile} | awk '{print $1}');
+            if [ $readlength -lt 28 ]; then readlength=28; fi
             echo "Metrics file with readlength " $readlength
         else
             readlength=~{read_length}
         fi
 
+        echo $readlength
         if [ -f "~{fastqfile_R2}" ]; then
             bowtie \
                 --chunkmbs=256 \
+                -l $readlength \
                 -p ~{ncpu} \
                 -k ~{good_alignments} \
                 -m ~{limit_alignments} \

diff --git a/workflows/tasks/macs.wdl b/workflows/tasks/macs.wdl
@@ -6,7 +6,7 @@ task macs {
         File bamfile
         File? control
 
-        Int memory_gb = 20
+        Int memory_gb = 25
         Int max_retries = 1
         Int ncpu = 1
 

diff --git a/workflows/tasks/peaseq_util.wdl b/workflows/tasks/peaseq_util.wdl
@@ -16,7 +16,7 @@ task fraggraph {
         String bam_location = "BAM_files"
         String annotation_location = "Annotation"
 
-        Int memory_gb = 50
+        Int memory_gb = 75
         Int max_retries = 1
         Int ncpu = 1
     }

diff --git a/workflows/tasks/sortbed.wdl b/workflows/tasks/sortbed.wdl
@@ -6,7 +6,7 @@ task sortbed {
 
         String outputfile = basename(bedfile,'.bed')+ '.sorted.bed'
 
-        Int memory_gb = 20
+        Int memory_gb = 30
         Int max_retries = 1
         Int ncpu = 1
     }

diff --git a/workflows/workflows/mapping.wdl b/workflows/workflows/mapping.wdl
@@ -20,7 +20,7 @@ workflow mapping {
         String? results_name
     }
 
-    if ( defined(metricsfile) ) {
+    if (! defined(fastqfile_R2) ) {
         call bowtie.bowtie as SE_map {
             input :
                 fastqfile=fastqfile,
@@ -35,6 +35,7 @@ workflow mapping {
             input :
                 fastqfile=fastqfile,
                 fastqfile_R2=fastqfile_R2,
+                metricsfile=metricsfile,
                 index_files=index_files,
                 insert_size=insert_size,
                 strandedness=strandedness,