Gwa Data Filtering

Question

[Sudeep]

I am doing a GWA study on pigs particularly bred for meat What are the data filtering approaches that can be used in GWAS study for these highly selected farm animals ? I know that filtering on HWE and PPC tests in plink will not be valid for these type of data with non random mating. For phenotype filtering I am doing and IBM based clustering in plink, then mds scaling in R , and then I filter out those samples for which the scaling values do not reach a certain threshold (say I filter on absolute value >0.01 ). Is this a valid approach? or are there any other valid approaches?

I appreciate your help.

[Istvan Albert]

Your process sounds good. A GWAS study on pigs bread for meat is one of those studies that I never thought were ongoing but they sure do sound pretty interesting.

Answer 1

[Larry_Parnell]

We work with moderately sized human populations, say 800-1600 subjects, and this may give us some luxuries that you won't have. Your approach sounds good and contains things that we have done in our studies. We also filter on call rate and on minor allele frequency. Mendelian segregation is also important to us. The values for your filters on these items, if you choose to incorporate them, are certainly likely to be dependent on your set of samples/subjects. These are our major items.

[Sudeep]

Hi thank you for the response, we have gone through the QC step (partially), now looking forward to fix some issues with family assignment of data and choosing a good methodology for association testing (we cannot use normal methods, these samples are highly related as well)

[Larry_Parnell]

We have family structure for our human studies, but not so highly related. It might be worth to look at what is happening in terms of genetics, traits and marker analysis in race horses. I offer this in light of what is likely to be a fair amount of relatedness. Dog breeds have also gone through a significant genetic bottleneck and you may find some hints with these groups.