auto trim length

Kincekara · Nov 17, 2023 · e2154df · e2154df
1 parent 4d8f2c9
commit e2154df
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 3 deletions.
diff --git a/tasks/task_fastp.wdl b/tasks/task_fastp.wdl
@@ -63,11 +63,12 @@ command <<<
     # parse output
     jq -r '.summary.fastp_version' fastp.json > VERSION
     jq '.summary.before_filtering.total_bases' fastp.json > total_bases.txt
-    jq '.summary.after_filtering.total_bases' fastp.json >> total_bases.txt
+    jq '.summary.after_filtering.total_bases' fastp.json >> total_bases.txt  
     jq '.read1_before_filtering.total_reads' fastp.json > R1_READS    
     jq '.read2_before_filtering.total_reads' fastp.json > R2_READS
     jq '.summary.before_filtering.total_reads' fastp.json > TOTAL_READS
     jq '.summary.after_filtering.total_reads' fastp.json > TOTAL_READS_TRIM
+
 
     r1_q30=$(jq '.read1_before_filtering.q30_bases' fastp.json)
     r1_total=$(jq '.read1_before_filtering.total_bases' fastp.json)
@@ -77,6 +78,9 @@ command <<<
     r1_total_trim=$(jq '.read1_after_filtering.total_bases' fastp.json)
     r2_q30_trim=$(jq '.read2_after_filtering.q30_bases' fastp.json)
     r2_total_trim=$(jq '.read2_after_filtering.total_bases' fastp.json)
+
+    echo $(($(jq '.summary.before_filtering.read1_mean_length' fastp.json) + 1 )) > read_length.txt
+
 
     if [ "$r1_total_trim" -gt 0 ]
     then
@@ -107,6 +111,7 @@ output {
     Int total_reads_trim = read_int("TOTAL_READS_TRIM")
     Int r1_reads = read_int("R1_READS")
     Int r2_reads = read_int("R2_READS")
+    Int read_length = read_int("read_length.txt")
     Float r1_q30_raw = read_float("R1_Q30_RAW")
     Float r2_q30_raw = read_float("R2_Q30_RAW")
     Float r1_q30_trim = read_float("R1_Q30_TRIM")

diff --git a/tasks/task_spades.wdl b/tasks/task_spades.wdl
@@ -4,7 +4,7 @@ task spades_pe {
   input {
     File read1
     File read2
-    Int? contig_threshold = 300
+    Int contig_threshold
     String samplename
     String docker = "staphb/spades:3.15.5"
 
@@ -22,6 +22,7 @@ task spades_pe {
     mv out/scaffolds.fasta ~{samplename}_scaffolds.fasta
 
     # remove short contigs
+    echo "removing contigs shorter than ~{contig_threshold}"
     python <<CODE
     import re
     with open("~{samplename}_scaffolds.fasta", "r") as input:

diff --git a/workflows/wf_c-bird.wdl b/workflows/wf_c-bird.wdl
@@ -66,7 +66,8 @@ workflow cbird_workflow {
       input:
       samplename = samplename,
       read1 = assembly_prep.read1_clean_norm,
-      read2 = assembly_prep.read2_clean_norm       
+      read2 = assembly_prep.read2_clean_norm,
+      contig_threshold = fastp_trim.read_length * 2    
     }
 
     if ( target_genes_fasta != 'null' ) {