This is a test version of Biostars. For the public version, visit https://www.biostars.org.
awk command to extract deletions from a CIGAR string

Hi Biostars, I am reanalyzing some RNA-seq data and would like to identify reads with the longest deletions. I have extracted the CIGAR strings containing deletions from column 6 of my SAM files. However, I have a huge list of reads and I don't want to count the number of deletions by eyeballing through this huge list. I know I need an awk command to filter things here, but for now I can't put together a complex one for this. I would be grateful for any assistance. 

samtools view sample.sam | cut -f 6 | grep "D" | head -3
12M1D89M
12M1D89M
13M1D88M
rna-seq alignment
• 3,849 views • link
updated by Pierre Lindenbaum Pierre Lindenbaum • written by Baby Bioinformatician Baby Bioinformatician

2 answers

You could do that with grep like so

echo 12M13D89M35D45M2I34M | grep -oP "(\d+)D" | tr -d "D"

will print

13
35
• 0 views • link
written by Istvan Albert Istvan Albert

using bioalcidaejdk: http://lindenb.github.io/jvarkit/BioAlcidaeJdk.html

$ java -jar  dist/bioalcidaejdk.jar  -e 'stream().filter(R->!R.getReadUnmappedFlag() && R.getCigar()!=null).filter(R->R.getCigar().getCigarElements().stream().anyMatch(C->C.getOperator()==CigarOperator.I || C.getOperator()==CigarOperator.N)).forEach(R->println(R.getCigar().getCigarElements().stream().filter(C->C.getOperator()==CigarOperator.I || C.getOperator()==CigarOperator.N).mapToInt(C->C.getLength()).max().orElse(0) +"\t"+ R.getSAMString()));'  toy.bam |\
sort -t $'\t' -k1,1n | tail | cut -f 2- 


r002    0   ref 9   30  1S2I6M1P1I1P1I4M2I  *   0   0   AAAAGATAAGGGATAAA   *   RG:Z:gid1
r001    163 ref 7   30  8M4I4M1D3M  =   37  39  TTAGATAAAGAGGATACTG *   RG:Z:gid1   XX:B:S,12561,2,20,112
x3  0   ref2    6   30  9M4I13M *   0   0   TTATAAAACAAATAATTAAGTCTACA  ??????????????????????????  RG:Z:gid1
r004    0   ref 16  30  6M14N1I5M   *   0   0   ATAGCTCTCAGC    *   RG:Z:gid1
• 0 views • link
written by Pierre Lindenbaum Pierre Lindenbaum

Log in to answer this question.

More posts like this