pindel data least fold needed

[jane_orderly]

reply from the author of pindel

hi Jane,

please post your questions to seqanswers or biostar in the future so that the discussion might be seen by others.

for Pindel, the more coverage the better sensitivity. pindel will find variants at low coverage as low as 4x as in 1000G but then we need to pool samples together. all depending the percentage of variants you want to discover, 10x for single sample will give you reasonable result and 20x above will be much better.

as for repetitive sequences, the reads have to be longer than the repetitive sequence to allow sensitive discovery. so again, the longer the better.

let me know if you have any questions. I could not give precise numbers as we normally work on high coverage data in cancer samples for somatic variants or low coverage pooled samples.

Kai

[jane_orderly]

Hello, I am using pindel to find deletions. First, I use real data(read length is 75bp) of 10 fold of a genome in which 85% are repetitive sequences, and i find some deletions.Then, simulated data were generated to evaluate the accuracy based on a real data of 200kb.I simulated 75bp,10 fold, 20 fold,50 fold respectively, at different fold, i found different numbers of deletion of 1,13(of which, 1 is not in the right chromosome), 47（of which, 5 is not in the right chromosome ）. Then my question is what do you think the least fold when using pindel to find deletion.
I saw published paper using 20X, 40X of 35bp read length,does it ok to use data of 10X at a read length of 75bp for a genome possessing 85% repetitive sequences？
Expecting to your reply!
Thanks!

jane