How to get ungapped single-end reads from the STAR output bam file?

Question

[caggtaagtat]

Hello,

After mapping with STAR, I would like to remove all reads (single-end) from my Bam file, which show a gap in their alignment. Can this be done with samtools somehow?

[Devon Ryan]

Are deletions OK or do you want them filtered too?

[caggtaagtat]

I would like to remove deletions as well

[Pierre Lindenbaum]

GAP = single nucleotide variation + indels ?

[caggtaagtat]

I mean the remaining reads should be only continiously mapped to the reference genome, without a gap in their alignment, through splicing or deletions. The occurrence of SNPs was allowed during mapping and should be disregarded at this step.

Edit: Sry, with gap, I meant a splice junction in the reads alignment

Answer 1

[Pierre Lindenbaum]

using samjdk: http://lindenb.github.io/jvarkit/SamJdk.html

$ java -jar dist/samjdk.jar -e 'return !record.getReadUnmappedFlag() && record.getCigar()!=null && record.getCigar().getCigarElements().stream().map(C->C.getOperator()).noneMatch(OP->OP.equals(CigarOperator.N) || OP.equals(CigarOperator.D));'  in.bam

• !record.getReadUnmappedFlag() read must be mapped
• record.getCigar()!=null read has cigar
• record.getCigar().getCigarElements().stream(). get a stream of cigar componenets
• map(C->C.getOperator()) map to the cigar operator
• noneMatch(OP->OP.equals(CigarOperator.N) || OP.equals(CigarOperator.D)); no operator can be 'N' or 'D'

[caggtaagtat]

Thank you very much! This is a very interesting tool and it seems to work.

Edit: It definitly worked right away