Ideally, no batch effect will be present because individuals with different phenotypes, sex, race, and other confounders should be plated randomly, and because modern high-throughput genotyping technology is much more accurate, efficient, and consistent than earlier generations of GWAS assays. There are several approaches for examining a dataset for potential batch effects. One simple approach is to calculate the average minor allele frequency and average genotyping call rate across all SNPs for each plate. Gross differences in either of these on any plates can easily be identified. Another method involves coding case/control status by plate followed by running the GWAS analysis testing each plate against all other plates. For example, the status of all samples on plate or batch 1 will be coded as case, while the status of every other sample is to be coded control. A GWAS analysis is to be performed (e.g. using the --assoc option in PLINK), and both the average p-value and the number of results significant at a certain threshold (e.g. p<1×10−4) can be recorded. SNPs with low minor allele frequency (i.e. <5%) should
