search

pcarbo / varbvs

Large-scale Bayesian variable selection for R and MATLAB.
http://pcarbo.github.io/varbvs
GNU General Public License v3.0
42 stars 14 forks source link

Do you recommend pruning SNPs that are in linkage disequilibrium before putting into varbvs? #34

Open garyzhubc opened 3 years ago

garyzhubc commented 3 years ago

Do you recommend pruning SNPs that are in linkage disequilibrium before feeding into varbvs? It looks like paper suggests that the variational approximation works better under independence of the SNPs? The problem with pruning is that it may prune away the effective SNPs while leaving the ones that are merely in linkage disequilibrium with the effective ones.

pcarbo commented 3 years ago

The problem with pruning is that it may prune away the effective SNPs while leaving the ones that are merely in linkage disequilibrium with the effective ones.

Precisely. So if that is a concern, you should not prune SNPs.

garyzhubc commented 3 years ago

But would that come at a cost that varbvs approximation becomes inaccurate?

pcarbo commented 3 years ago

Yes, although despite this inaccuracy, some posterior statistics remain accurate, such as aggregate statistics over all SNPs in a chromosomal region; see the Bayesian Analysis paper for more discussion on this.