Genotype inference allows us to evaluate the evidence for association at the genetic markers that are not directly-genotyped. It helps improve the power of individual scans and is of particular usefulness for combining information from different studies such as linkage and GWAS. However, the accuracy of genotype inference may be impaired for the following reasons. First, because datasets where subjects are genotyped on different platforms may have different genotyping error rates, when we combine these datasets, inference can be problematic. Second, genotype inference for large datasets based on a small amount of shared information may encounter too much uncertainty in the procedure. For a similar reason, SNPs with low MAF may also have a higher chance of being inferred inaccurately.
