vince
New Member
Posts: 6
|
Post by vince on May 1, 2019 0:10:17 GMT
Hi,
I am keen to use fastGWA for UKB data. Thanks for developing this tool.
Does fastGWA support BGEN format? If not, how did you perform the GWAS of UKB imputed data?
Thanks,
Vince
|
|
vince
New Member
Posts: 6
|
Post by vince on May 1, 2019 14:15:40 GMT
After reading the bioarxiv paper, www.biorxiv.org/content/biorxiv/early/2019/04/11/598110.full.pdf, I now see that: 1. BGEN was converted to PLINK binary format and only for SNPs with MAF > 0.0001 and INFO > 0.8. 2. That the full GRM was likely computed in parts followed passing each part into making the sparse GRM, finally merging the parts the sparse GRM into a single sparse GRM. Is the above correct? Thanks, Vince
|
|
vince
New Member
Posts: 6
|
Post by vince on May 8, 2019 14:02:33 GMT
To close this question here is what I ended up doing according to email correspondence with fastGWA authors:
1. BGEN was converted to PLINK binary format and only for SNPs with MAF > 0.0001 and INFO > 0.8. 2. Compute full GRM in parts, merge parts to single full GRM, then convert single full GRM to spare GRM.
Resulting assoc stats where compared to those from BOLT-LMM and found to be highly concordant.
V
|
|