Computing time #65
Description
Hello,
I use BGLR for genomic predictions and I have very long computing times (days) with a dataset comprising 22K SNPs and 200K data records, and a GBLUP approach. I think this is mostly because the function takes a lot of time before starting the iterations.
I thought it was normal, but I saw in your article (Figure S7 of Fine mapping and accurate prediction of complex traits using Bayesian Variable Selection models applied to biobank-size data) that you had much lower computing times with 10K SNPs and 300K samples (<30minutes). What could be the reason for that ? I use a high performance computing cluster, the BGLR function and 40K iterations. My model has 3 fixed effects, 3 random effects (iid) and 1 random effect associated with the G matrix.
Thanks
David