同步至HPC-v20190527版

tasks/TrimAdapt.wdl

 task TrimAdapt {
-    String sample_ID
+    String sample_id
     File in_fastq
 
     String adapter_seq
     Int randomBase_in_adapter
+    Int sequencing_length
 
     String docker
     String cluster_config
         set -e
         nt=$(nproc)
         
-        echo -e "Trim adapter's invariant part." > ${sample_ID}.trimAdapt.log
-        fastp --thread $nt -Q -L\
+        echo -e "Trim adapter's invariant part." > ${sample_id}.trimAdapt.log
+        
+        fastp --thread $nt -Q \
+            --length_required 0 \
+            --length_limit $[${sequencing_length}-1] \
             --adapter_sequence ${adapter_seq} \
             -i ${in_fastq} \
-            -o ${sample_ID}.trimAdapt.fastq.tmp.gz \
-            2>> ${sample_ID}.trimAdapt.log
+            -o ${sample_id}.trimAdapt.fastq.tmp.gz \
+            2>> ${sample_id}.trimAdapt.log
 
         if [ ${randomBase_in_adapter} -gt 0 ]
         then
-            echo -e "\nTrim ${randomBase_in_adapter} random base from both sides\n" >> ${sample_ID}.trimAdapt.log
+            echo -e "\nTrim ${randomBase_in_adapter} random base from both sides\n" >> ${sample_id}.trimAdapt.log
             fastp --thread $nt -A -Q -L \
                 --trim_front1 ${randomBase_in_adapter} --trim_tail1 ${randomBase_in_adapter} \
-                -i ${sample_ID}.trimAdapt.fastq.tmp.gz \
-                -o ${sample_ID}.trimAdapt.fastq.gz \
-                2>> ${sample_ID}.trimAdapt.log
+                -i ${sample_id}.trimAdapt.fastq.tmp.gz \
+                -o ${sample_id}.trimAdapt.fastq.gz \
+                2>> ${sample_id}.trimAdapt.log
         else
-            mv ${sample_ID}.trimAdapt.fastq.tmp.gz ${sample_ID}.trimAdapt.fastq.gz
+            mv ${sample_id}.trimAdapt.fastq.tmp.gz ${sample_id}.trimAdapt.fastq.gz
         fi
 
-        zcat ${sample_ID}.trimAdapt.fastq.gz | paste - - - - | cut -f 2 | \
+        echo -e "Length\tReadCount" > ${sample_id}.trimAdapt.lengthDistribute
+        zcat ${sample_id}.trimAdapt.fastq.gz | paste - - - - | cut -f 2 | \
             awk '{a[length($1)]++}END{for(i in a){print i,a[i]}}' | sort -n \
-            > ${sample_ID}.trimAdapt.lengthDistribute
+            >> ${sample_id}.trimAdapt.lengthDistribute
 
     >>>
 
     }
 
     output {
-        File out_fastq="${sample_ID}.trimAdapt.fastq.gz"
-        File out_log="${sample_ID}.trimAdapt.log"
-        File out_lengthDistribute="${sample_ID}.trimAdapt.lengthDistribute"
+        File out_fastq="${sample_id}.trimAdapt.fastq.gz"
+        File out_log="${sample_id}.trimAdapt.log"
+        File out_lengthDistribute="${sample_id}.trimAdapt.lengthDistribute"
     }
 }