Set: Choose a to minimize and now calculate X = RR′ instead of MM′. It is first important to check that in the absence of LD the suggested correction does not seriously distort the Tracy–Widom statistic. In Figure 9A and 9B we show P–P plots, uncorrected, and with five levels (k = 1…5) of correction. The first figure is with 100 individuals and 5,000 markers, the second with 200 individuals and 50,000 markers. Then in Figure 10A and 10B we analyze a simulated dataset with severe LD. We generate blocks in perfect LD, in which the probability that a block contains L markers is 2−L. We show the corresponding plots. Note that here the uncorrected statistic is distributed quite differently than the Tracy–Widom distribution. Our suggested correction strategy seems to work well, and should be adequate in practice, especially as most large genotype arrays will attempt to avoid high levels of LD. We would recommend that before analyzing a very large dataset with dense genotyping, one should filter the data by removing a marker from every pair of markers that are in tight LD.
