the quality of novel SNPs

[caswater]

I processed my file, and the WGS result showed around 30k novel SNPs that have not been annotated in dbSNP135. However, the Ti/Tv ratio for all SNPs is around 2.11, which is expected from a WGS experiment; however the novel SNPs only have Ti/Tv around 1.46, indicating low specificity...

I am pretty interested in some novel SNPs, but they show overall low Ti/Tv ratio... how shall I pick up some confident novel SNPs?

Thanks for your advice,

[aeonsim]

This is really a question everyone would like to have answered, and to which (I at least am aware of no bioinformatic approach) there is no simple method. It depends on the biology of the organism and the purpose to which you wish to use the data. Generally you need to decide on some biological relevant limits and filter on those standard ones are:
Read depth, quality score, gq score, allele ratio, hwe and so on but these are all dependent on your data set, goal and pipeline.

One bioinformatic thing you can do is run a couple more pipelines ie rtg, freebayes or Samtools merge the results and look at novel variants that have been called by all three pipelines. Or if you have a pedigree/trio use Mendelian rules to narrow it down.

At the end of the day though if you want to do anything with them or use them in a published study you'll need to go back to the lab and validate the variants with different genotyping technologies, Sanger sequencing, sequenom iplex or some other of the many methods.

[HESmith]

I'll second aeonsim's point re: using the intersection of multiple pipelines to identify high-confidence SNPs. We've compared the results of MAQ and BFAST/SAMTools, and found that >80% of the SNPs called by both are validated by Sanger sequencing (YMMV).

[caswater]

Thanks a lot, and your suggestion is really helpful