Hi,

I'm about to do two-sample mendelian randomization analysis and would like to exclude weak instrument bias by calculating F statistic.

The following formula is commonly used to calculate F statistic;

n=sample size; k=number of instruments; R2=the proportion of variance in the phenotype explained by the genetic variants

How would this R2 approximate from the information in the GWAS summary statistic?

Also, I assume that logistic regression is used to calculate GWAS when using a binary trait such as presence of disease as exposure, can that formula be used as well?

Your help will be appreciated in this matter, Thank you.