What are the best HW p-value & MAF cutoffs based on my study ?

Question

[mona.patoughi]

Hello,

I understand that my question in not completely related to bioinformatics but I would like to read the experts comments who are working with genomics data :

Based on my research question, I would like not to exclude variations that are out of Hardy-Weinberg equilibrium, and I also want to have rare variants in my study ( i mean the MAF threshold must be include rare variants). are these cutoffs appropriate for my analysis:

• **HW P-values = 0.001
• MAF = 0.001**

I appreciate your help.

Answer 1

[LChart]

The Hardy-Weinberg exclusion is generally there to remove poor-quality or artefactual variants. While population-stratified variants will also be excluded, population stratification is handled statistically by mixed effects.

My recommendation is to take a look at the variants that are out of HWE (p < 1e-4 or so), and go back to any "raw" data you may have (intensity plots for chips, INFO statistics for sequencing) and see if any variants have clear hallmarks of some kind of calling artefact. You can use these hallmarks to define surrogate thresholds to deplete artefacts from out-of-HWE variants; and you could even go further to define a small set of true-positives and false-positives to build a classifier.

If you don't have any raw data, then one of the simplest things you can do is fall back to genome in a bottle and stratify your analysis by "variant in a difficult region" or "variant not in a difficult region": https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/release/genome-stratifications/v3.3/GRCh38@all/Union/

[mona.patoughi]

Thanks a lot for your fantastic answer.

I have the raw data. However, based on this sentence : "and see if any variants have clear hallmarks of some kind of calling artefact." I dont know how I can recognize the clear hallmark variants !

appreciate your nice consideraton.

Regards, Mona

[LChart]

Look at the reads in IGV; see https://bioinformatics-core-shared-training.github.io/intro-to-IGV/InspectingVariantsInIGV.html

[mona.patoughi]

Appreciate your guide.

Regards, Mona