Scaling of computational time

[AurinaBMH]

Hello,

I am planning a large-scale neuroimaging analysis that would also require extracting individual univariate GWAS results for each phenotype. Would you be able to advise on how the computational time for mostest scales with the number of phenotypes, e.g. 5000 vs 10 000 phenotypes for 40k subjects (~10mln SNPs)? Or alternative software for univariate GWAS would be advised in this case?

Thank you for your help.

Best,
Aurina

[baihongguo]

Hi Aurina,

I feel MOSTest is not efficient in getting univariate GWAS results because of the permutation. I would recommend fastGWA or plink2 for univariate GWAS, it's much faster. And FYI: An efficient multivariate GWAS tool taking univariate GWAS summary statistics: https://github.com/hanchenlab/JAGWAS. If you want to jointly analyze your high-dimensional neuroimaging phenotypes.

Sincerely,
Bohong