TP métagénomique

16S_workflow.sh 2.0KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758
  1. #!/bin/bash
  2. raw_reads_dir=$1
  3. output_dir=$2
  4. mkdir $output_dir
  5. mkdir trimmed_reads
  6. #TRIM READS WITH ALIENTRIMMER
  7. for R1 in $(ls $raw_reads_dir/*R1.fastq);
  8. do
  9. R2=`echo $R1 | sed 's/R1/R2/g'`
  10. oR1=`echo $R1 | sed 's/\.fastq/\.at\.fq/g'`
  11. oR1=`echo $oR1 | sed 's/fastq\///g'`
  12. oR2=`echo $R2 | sed 's/\.fastq/\.at\.fq/g'`
  13. oR2=`echo $oR2 | sed 's/fastq\///g'`
  14. java -jar soft/AlienTrimmer.jar -if $R1 -ir $R2 -q 20 -c databases/contaminants.fasta -of trimmed_reads/$oR1 -or trimmed_reads/$oR2
  15. done
  16. # MERGE READS AND OUTPUT FASTA
  17. mkdir fasta
  18. for R1 in $(ls trimmed_reads/*R1.at.fq);
  19. do
  20. R2=`echo $R1 | sed 's/R1/R2/g'`
  21. outname=`echo $R1 | sed 's/_R1\.at\.fq/\.fasta/g'`
  22. label=`echo $outname | sed 's/trimmed_reads\///g'`
  23. outname=`echo $outname | sed 's/trimmed_reads/fasta/g'`
  24. label=`echo $label | sed 's/\.fasta//g'`
  25. soft/vsearch --fastq_mergepairs $R1 --reverse $R2 --fastaout $outname --label_suffix ";sample=$label;"
  26. done
  27. # MERGE ALL FASTAS TO 'amplicon.fasta'
  28. cat fasta/*.fasta > fasta/amplicon.fasta
  29. # REMOVE ' ' FROM EACH READ NAME
  30. sed -i 's/ //g' fasta/amplicon.fasta
  31. # CREATE OTU
  32. # DEDUPLICATION
  33. mkdir otu
  34. soft/vsearch --derep_fulllength fasta/amplicon.fasta --output fasta/fulllength.fasta
  35. soft/vsearch --derep_prefix fasta/amplicon.fasta --output fasta/prefix.fasta
  36. # REMOVING SINGLETON
  37. soft/vsearch --derep_prefix fasta/amplicon.fasta --output fasta/prefix_min10.fasta --minuniquesize 10
  38. # REMOVING CHIMERAS
  39. soft/vsearch --uchime_denovo fasta/prefix.fasta --nonchimeras fasta/prefix_nochim.fasta
  40. # CLUSTERING
  41. otu=">OTU_"
  42. vsearch --cluster_fast fasta/prefix_nochim.fasta --id 0.97 --centroids otu/otu.fasta --relabel $otu
  43. # ABONDANCE
  44. vsearch --usearch_global fasta/amplicon.fasta --db otu/otu.fasta --id 0.97 --otutabout otu/abondance_table.txt
  45. # ANNOTATION
  46. soft/vsearch --usearch_global otu/otu.fasta --db databases/mock_16S_18S.fasta --id 0.97 --top_hits_only --userfields query+target --userout otu/annotation.txt