search

rondolab / MR-PRESSO

Performs the Mendelian Randomization Pleiotropy RESidual Sum and Outlier (MR-PRESSO) method.
81 stars 30 forks source link

Question about global p-value #20

Open ruislab opened 1 year ago

ruislab commented 1 year ago

Hi,

Thanks for making this useful package for detecting outliers and pleiotropy.

Recently, I am running an MR analysis, and doing sensitivity analysis using MRPRESSO. However, one of my results shows that the global P-value is 1. I am wondering whether this situation is due to a very close to 1 p-value, which is displayed as 1, or the p-value is exactly 1 because of the data property.

marieverbanck commented 1 year ago

Hi @ruislab, thanks for using MR-PRESSO. Indeed, please see reply to issue #8. Since empirical P-values are computed, if you wish to increase the precision in the estimation of your p-values you should increase the number of permutations specified in NbDistribution. Best, Marie