GCTA-COJO

new

« Prev
1
Next »

joseph
New Member

Posts: 1

GCTA-COJO Nov 25, 2014 17:43:25 GMT

Quote

Post by joseph on Nov 25, 2014 17:43:25 GMT

Hi,

I've run GCTA-COJO using:
--bfile bfile
--massoc-file file
--massoc-joint
--out outfile

I can find in log file the following:
"Phenotypic variance estimated from summary statistics of all 67 SNPs: 124.379 (variance of logit for case-control studies)."

Can I determine from this value the percentage of variance explained by those 67 SNPs? Please note that individual-level genotype data of the discovery set is not available.

Many thanks

Zhihong Zhu
Moderator

Posts: 88

GCTA-COJO Nov 27, 2014 7:56:57 GMT

Quote

Post by Zhihong Zhu on Nov 27, 2014 7:56:57 GMT

Hi Joseph,

I'm afraid I don't think the SNP heritability could be estimated from summary data. Because the "phenotypic variance" estimated from summary data and reference genotypes seems like predicted variance, which is related to the SNP heritability and sample size. Under the assumption that the reference sample and the discovery sample (to estimate beta, se...) have the same ancestry, similar allele frequency, I think it's fine to re-estimate beta and se in joint analysis.

Cheers,
Zhihong