How do I find medium length somatic indels?

[lh3]

I tried "bwa aln -e 50" and it seems to work. Is bwa_mem better than bwa for this? How so?

Yes, that may work, but bwa-mem should be better, at least faster. Bwa and bwa-mem are completely different algorithms.

[ymc]

I tried "bwa aln -e 50" and it seems to work. Is bwa_mem better than bwa for this? How so?

[KatsenPlatz]

Strelka does a fantastic job of finding longer somatic indels: https://sites.google.com/site/strelk...variantcaller/

For exome data, set the "isSkipDepthFilters" to 1 in order to skip depth filtration.

[lh3]

For single-end reads, try novoalign. You may also consider bwa-mem with (higher matching score):

bwa mem -A2 -B3 -O5 -E1

bwa-backtrack does not find a gap more than 4bp for single-end reads. Pindel would not work with single-end, as I remember (could be wrong).

[KaiYe]

try Pindel. there is no theoretical maximum deletion size. currently maximum insertion size for insertions is read length - 20.

[ymc]

I used IndelGenotyperV2 to find somatic indels. But the longest I find is 4bp. My reads are single 75bp exome reads. I am interested in indels around 20bp. Is that possible with my reads?

I ran the mapping step with bwa-0.7.5 on default setting. Is that the cause? I heard that the -o and -e flags for bwa aln supposedly have something to do with indels. If so, how do I set it up to find indels around 20bp?

Thanks a lot in advance!