phenotypic correlation for step 3
nrs225 opened this issue · 3 comments
Hi,
Thank you for developing this tool.
I am using the local genetic correlation section of the tool (rho-HESS) and have a question regarding accounting for phenotypic correlation which is required to be fed in to step 3.
In the manual (https://huwenboshi.github.io/hess-0.5/local_rhog/), it states that the genetic covariance intercept term from cross-trait LDSC provides an approximation of phenotypic correlation. As I am using publicly available GWAS data from different consortia which consist of multiple source studies, possibly including UK Biobank, I am unsure about the number of overlapping samples between the 2 input GWAS datasets.
Should I not use this LDSC intercept value instead of calculating the exact phenotypic correlation (for which I do not have exact numbers to run), or should I use phenotypic correlation = 0 (as used in the rho-HESS publication) purely because they aren't identical consortia?
Many thanks in advance!
Hi,
I have the same issue.Have you solved it yet?
Many thanks in advance!
Hi @yangrousong, I have not solved the issue and applied the flags --num-shared 0 and --pheno-cor 0 since my input GWAS summary statistics were from different consortia.
Thank you for the response. However, I have some summary statistics with overlapping samples. I'm going to keep thinking about it. Thank you again.