Variant callers for Ion Torrent data?

Question

[Jordan]

Hi,

What do people use for variant calling on Ion torrent data? I tried germline variant calling from varscan2, but there seem to be too many variants reported, more specifically indels. I realize this is because of the homopolymer sequencing issue with Ion torrent.

Are there tools available for taking into account the poor homopolymers sequencing? Any help is appreciated.

Thanks!

Answer 1

[Charles Warden]

I've found calling Ion Torrent to be awkward. The machine probably already provided .vcf files for each sample, I would would recommend using those (which have a platform-specific error model). However, you will probably want to take the overlap of multiple variant lists (such as the default variant list and your variant calling list). The default list will be for a single sample, but I believe Torrent Suite also has a somatic variant caller (if you want to try that - I haven't, so i can't provide more specific advice).

[Jordan]

Thanks Charles. I think we have the vcf files provided from Ion Torrent.

I'm not sure comparing my variant calls with ion torrent vcf files would help. There are too many false positives in my vcf files. I will try torrent suite.

Answer 2

[Robert Sicko]

Have a look at a related question here. Also, there are a ton of parameters for variant calling right in torrent server; perhaps you can ask whom ever ran the samples to call variants on torrent server using parameters you specify, or you can install stand alone torrent variant caller (TVC) off iontorrent's github. One final alternative is to use ion reporter which allows up to 100GB of data free (I believe).

[Jordan]

I didn't realize we had stand alone variant caller from ion torrent. I will try using that. Thanks for the help!

[Greg Tyrelle]

The github source is not what you want for the stand-alone variant caller. See this page on the Ion Community:

http://ioncommunity.lifetechnologies.com/docs/DOC-8932