Is Post Filtering Of Vqsr Calls Necessary With Gatk?

Question

[User 1666]

Hi,

I'm using GATK to analyze whole exome experiment data

After applying Variant Recalibration, is it necessary to perform some additional filtering based on HaplotypeScore, MQ, QD etc.? As I understood, Variant Recalibration is an alternative to hard filtering.

Thanks in advance,

Andrei

[Khader Shameer]

Can you tell us a bit more about your analysis steps: including version of GATK and walkers used in recalibration step (CountCovariates + TableRecalibration or BaseRecalibrator + PrintReads) ?

[User 1666]

The whole pipeline we are using generally complies with what GATK people recommend http://www.broadinstitute.org/gatk/guide/topic?name=intro

Answer 1

[Omid]

As far as I know Variant Recalibration is alternative to hard filtration and I think better than hard filtration. Because in VQSR the software will consider all criteria in comparison to sequence data with a good quality.Then based on VQSLOD you decide about your variant. I think Variant Recalibration is better because based on just one item(VQSLOD) you can decide about your variant. But in hard filtration we will have several criteria like MQ0,Hrun,FS, haplotyescore... it might be in some criteria the variants meet the threshold but in the others not, in this situation it is difficult to decide about variant. As far as I know the sample size is important when you want to apply Variant Recalibration but I do not know how many samples are enough for Variant Recalibration.If you found the answer please let me know.

[Omid]

May be my last question and its answer is useful to you http://www.biostars.org/post/show/49453/filtering-strategy-in-exome-sequencing-and-quality-control/ If you would like to know the thresholds let me know.