dan
New Member
Posts: 1
|
Post by dan on Sept 30, 2013 15:25:02 GMT
Hi everybody,
In the estimation of heritability, the standard deviation of h^2 seems to be related on the sample size. Is it also related to the number of SNPs? If we have a sample size of 1,500 individuals and 500,000 SNPs, the sandard deviation of h^2 is the same as the one estimated with the same sample size and 100,000 SNPs?
Thanks for your help,
Dan
|
|
|
Post by Jian Yang on Oct 7, 2013 14:39:36 GMT
Yes, standard error is smaller for smaller number of SNPs. You can test that empirically.
|
|
|
Post by Anna on Oct 8, 2013 14:30:57 GMT
Good afternoon, May I ask you a stupid question please? I have build the file *.ma. It looks like a file from your description, but i have an error "Invalid sample size in line". Can you recommend a way for creating .ma files please? Thank you.
|
|
|
Post by Jian Yang on Oct 9, 2013 13:13:15 GMT
I usually just read the meta-analysis summary data file in R and save the corresponding columns to the *.ma file. Could you please check a few lines using a text editor?
|
|
|
Post by SM on Nov 8, 2013 15:30:48 GMT
I am doing some multivariate fusion analysis with Imaging and genetic data, so our sample sizes are not that large compared to conventional GWAS type approaches. Would it be proper to estimate heritability using GCTA using about 600 subjects and about 20k SNPs?
|
|
|
Post by Jian Yang on Nov 10, 2013 8:28:37 GMT
SE is ~300/N with N being the sample size. Given a sample size of ~600, I would expect the SE being ~0.5 which is too large.
|
|
|
Post by Shareefa on Dec 4, 2013 8:25:20 GMT
Hi
I've got a sample size of 794 and about 300K SNPs. Regardless of which parameters I change, I always get a V(G)= o.oooooo. Does this indicate my sample size is too small?
Thanks!
|
|
|
Post by Jian Yang on Dec 4, 2013 10:45:17 GMT
Yes, sample size is too small. My guess is that your SE would be ~0.4.
|
|