mac76
New Member
Posts: 4
|
Post by mac76 on Jul 19, 2014 12:27:03 GMT
I've created GRM based on 1000 Genomes imputed data. Does anyone have thoughts on the impact that that many SNPs in LD could have on point estimates? Should the LD be a problem for my estimates? Would it be safer to prune data before calculating GRM? I used 1KG data as was hoping that rarer tags (although in LD with the common SNPs) could provide more power for tagging underlying rarer causal SNPs. From the latest publications by Jian I understand that the best way forward is to 'bin' GRM by MAF frequency bands, is that still good for 1KG? Thanks a lot for any feedback!
|
|