SNP and indel calling methods

[Rachelly]

Dear all,

I'm wondering how to do a good SNP and indel calling. There are a lot of aligners each contains different parameters and gives different results.
There is always the dilemma between results that is significant but partial and results that are wide but contain false-positives.

In 1000-genome project site they say about SNP calling methods:
"We found empirically that, given the state of current methods, the consensus of multiple primary call sets from different methods proved of higher quality than any of the primary call sets themselves. ... Overall, genotypes obtained through a consensus procedure are estimated to have 30% fewer errors than those generated by any single caller."

- I wonder if there's a tool that does alignments with different parameters/ tools and then merges the results using statistical methods?

- What would you say are the best SNP calling tools? (bowtie? GATK? samtools? others?...)

Thanks,
Rachelly.

[colindaven]

I haven't heard of a consensus SNP aligner which parses the output from multiple SNP callers, you'll probably not find this too tricky to do yourself.

We have had good results with the VarScan program, but have yet to really test the predictions in the wetlab. Previous results from the ELAND aligner and custom perl scripts using quality sums weren't too great to validate, that is, there were a lot of surprising false positives despite high-ish coverage (10+).