Nonfounder Ld Pruning

Question

[Sudeep]

I have a gwa data set with genotyping data only for nonfounders in the population Can any one suggest LD pruning methods for this data set? or can I assume that the samples in the data set are all founders and proceed with normal LD pruning? Thanks

[Michael]

let's face it, this is a tough question for most of us especially without too much detail. Can you help the few population geneticists (is that the correct term?) around by providing some detail about the type of data (genotyping platform), software you are using and the population? Further, have you ever seen a connection between (non)-founders and LD pruning been made in the literature, I couldn't find anything that would give a clue how these could be connected.

[David Quigley]

Do you have reason to think any of the nonfounders are siblings?

[Sudeep]

@David some of the samples in the population are full siblings , some are half siblings and the rest are unrelated

[Michael]

I guess you are using Plink and ask for an optimal set of parameters for its LD-pruning method?

Answer 1

[Jarretinha]

Being founder or nonfounder isn't very important. The main question must be: for how long did the population mix (in generations)? As LD decays exponentially with time, most loci normally will be free of significant LD in less than a dozen generations unless selection/migration/demographic stochasticity/low recomb have a relevant role in your scenario. So, check the age of your population, if it's young proceed with LD pruning carefully (they're just like founders).

By the way, the founder effect is relevant only when effective population size is small. For most vertebrates (except some fishes, maybe) this is always the case. So, a priori, all vertebrates are founders (conservatively).

Answer 2

[Genotepes]

Really depends on what you want to do with. If you want to do SNPs selection in LE for linkage, then take the same SNPs in 1000G and do the pruning there. You have more people and, moreover, if there is a "founder" effect in your pedigree which results in a long haplotype, you'd avoid breaking this very effect you are looking for by selecting only one SNP in this "block"