detect large indels

Question

[nakanomasayuki265]

What is the best way to detect large indels ? I mapped reads of whole genome shotgun sequencing against the reference sequences.

Answer 1

[Brian Bushnell]

The best way to detect large deletions is to map with BBMap and call variants with the BBMap package's CallVariants tool. BBMap can also detect moderate-sized insertions (up to 60% of insert size, or so). For example:

bbmap.sh in=reads.fq out=mapped.sam ref=ref.fa maxindel=400k
callvariants.sh in=mapped.sam out=vars.vcf ref=ref.fa ploidy=2

For longer insertions that cannot be captured within a read pair it's probably better to use an indel-specific caller based on coverage and read pairing. But for insertions shorter than ~60% of insert size it's better to merge reads, map them, and call them with a normal variant-caller. I've used BBMerge -> BBMap -> CallVariants successfully with >200bp insertions from 2x100bp reads.

Note that I wrote BBMap and I am biased.

Answer 2

[2nelly]

Some good tools are:

https://github.com/dellytools/delly

https://github.com/zeeev/wham

https://github.com/arq5x/lumpy-sv