MR Dictionary

Unweighted regression

Similar to the IVW (fixed effects) in concept, but all SNPs are weighted equally. 

The intention of the method is that in some cases a GWAS will return one extremely large effect on the exposure and many small effects, and that single large effect will dominate the analysis. The method is seldom used in practice though, as alternative approaches such as leave-one-out analysis can be used for sensitivity instead. 

Other terms in 'Two-sample MR methods':