05_results_combine.sh 2.4 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152
  1. sample=$1
  2. indir=$2
  3. #reads dimer
  4. reads_file=$indir"/"$sample"/1.QC/"$sample"_reads.txt"
  5. total_count=`awk '{if(NR>1){print $0}}' $reads_file | wc -l`
  6. dimer_count=`grep dimer $reads_file | wc -l`
  7. dimer_rate=`awk -v x=$total_count -v y=$dimer_count 'BEGIN{printf "%.2f\n",y*100/x}'`
  8. #fusion count
  9. IPC_count_file=$indir"/"$sample"/2.mapping/"$sample".IPC.reads.txt"
  10. nonspecific_file=$indir"/"$sample"/2.mapping/"$sample".nonspecific.reads.txt"
  11. target_file=$indir"/"$sample"/2.mapping/"$sample".target.reads.txt"
  12. fusion_file=$indir"/"$sample"/4.fusion/fusion_results.txt"
  13. IPC_count=`cat $IPC_count_file | wc -l`
  14. nonspecific_count=`cat $nonspecific_file | wc -l`
  15. target_count=`cat $target_file | wc -l`
  16. fusion_count=`awk '{sum+=$NF}END{print sum}' $fusion_file`
  17. fusion_rate=`awk -v x=$total_count -v y=$fusion_count 'BEGIN{printf "%.2f\n",y*100/x}'`
  18. IPC_rate=`awk -v x=$total_count -v y=$IPC_count 'BEGIN{printf "%.2f\n",y*100/x}'`
  19. nonspecific_rate=`awk -v x=$total_count -v y=$nonspecific_count 'BEGIN{printf "%.2f\n",y*100/x}'`
  20. target_no_fusion_count=`expr $target_count - $fusion_count`
  21. target_no_fusion_count_rate=`awk -v x=$total_count -v y=$target_no_fusion_count 'BEGIN{printf "%.2f\n",y*100/x}'`
  22. target_rate=`awk -v x=$total_count -v y=$target_count 'BEGIN{printf "%.2f\n",y*100/x}'`
  23. Misc_count=`expr $total_count - $IPC_count - $nonspecific_count - $fusion_count - $dimer_count - $target_no_fusion_count`
  24. Misc_rate=`awk -v x=$total_count -v y=$Misc_count 'BEGIN{printf "%.2f\n",y*100/x}'`
  25. #outfile
  26. QC_stat_file=$indir"/"$sample"/"$sample"_QC_stat.txt"
  27. echo -e "Sample\tTotal\tDimer\tDimer_Rate(%)\tFusion\tFusion_Rate(%)\tIPC\tIPC_Rate(%)\tNonspecific\tNonspecific_Rate(%)\tMisc\tMisc_Rate\tTarget_no_fusion\tTarget_no_fusion_Rate(%)" > $QC_stat_file
  28. echo -e "$sample\t$total_count\t$dimer_count\t$dimer_rate\t$fusion_count\t$fusion_rate\t$IPC_count\t$IPC_rate\t$nonspecific_count\t$nonspecific_rate\t$Misc_count\t$Misc_rate\t$target_no_fusion_count\t$target_no_fusion_count_rate" >> $QC_stat_file
  29. #fusion counts
  30. fusion_stat_file=$indir"/"$sample"/"$sample"_fusion_stat.txt"
  31. awk '{if(NR==1){print "Sample\t"$0}}' $fusion_file > $fusion_stat_file
  32. awk '{if(NR>1){print "'$sample'\t"$0}}' $fusion_file >> $fusion_stat_file
  33. echo "Done!"
  34. #reads primer check
  35. primer1=$indir"/"$sample"/1.QC/"$sample"*reads_primer_check*"
  36. sam1=$indir"/"$sample"/2.mapping/"$sample"*.sam"
  37. pe_bam=$indir"/"$sample"/2.mapping/"$sample".pe.bam"
  38. rm $primer1 $sam1 $pe_bam
  39. echo "Remove Done!"