Upload New File

Filtre_CIGAR

+#!/bin/bash
+module purge
+
+#$1 pourcentage identité
+#$2 pourcentage de la longueur
+for i in */*_mapped_fusariose.sam
+do name=$(basename $i .sam)
+echo $name
+
+awk '$0~"^@"' $i > $name\_filtered_CIGAR_$1\_id_$2\_length_paired.sam
+
+#attention la sortie awk convertit les tab en espace, besoin d'un sed pour remplacer les espaces en tabulation
+
+awk '$0!~"^@"' $i | awk '{if($6!="*"){print $0}}' | awk '{a=$6;gsub("M","_M;",a);gsub("I","_I;",a);gsub("D","_D;",a);gsub("S","_S;",a);print $0,a}' | awk -v name=$name id=$1 len=$2'{split($NF,tab3,";");tab2["M
+"]=0;sum=0;for (i=1;i<=length(tab3);i++){split(tab3[i],tab,"_");sum=sum+tab[1];tab2[tab[2]]=tab2[tab[2]]+tab[1]};print tab2["M"]/sum >> name"_freq.txt";if(tab2["M"]/sum>id && sum > len){print $0}}' | awk '{$NF
+="";print $0}' | awk '$2==99 || $2==147' |sed 's/ /\t/g'>> $name\_filtered_CIGAR_$1\_id_$2\_length_paired.sam
+
+module load bioinfo/samtools-1.9
+samtools view -S -b $name\_filtered_CIGAR_$1\_id_$2\_length_paired.sam > $name\_filtered_CIGAR_$1\_id_$2\_length_paired.bam
+module unload bioinfo/samtools-1.9
+
+module load bioinfo/bedtools-2.27.1
+bamToFastq -i $name\_filtered_CIGAR_$1\_id_$2\_length_paired.bam -fq $name\_filtered_CIGAR_$1\_id_$2\_length_paired_R1.fq -fq2 $name\_filtered_CIGAR_$1\_id_$2\_length_paired_R2.fq
+module unload bioinfo/bedtools-2.27.1
+
+done
+
+gzip  *R1.fq *R2.fq