Mendelian randomization is gaining popularity for studying the causal effect of an exposure on an outcome. The two stage least squares (TSLS) regression is a useful technique and is typically approximated by the inverse variance weighting (IVW) method from meta analysis. This presentation serves two purposes. The first is to introduce some new developments in methodology, focusing on those I have created: 1) The TSLS estimate can be computed from GWAS summary statistics without using the IVW approximation. 2) The heterogeneity of two samples can be checked graphically. 3) In the presence of two heterogeneity samples, the MR-Egger regression can be used after adopting appropriate weights. The main purpose of this presentation is to generate some interest, either in research or in application, on this emerging field. Some potential research topics are discussed.