A refined map of ALDH2*504Lys allele frequency was generated from these frequency data using the Surfer 8.0 program to interpolate the clinal patterns (Figure 1). For some of the populations, more than one sample was studied. The different samples from the same population usually had similar allele frequencies, while some showed notable deviation from the common data. Some of this inconsistency may have resulted from technical problems such as typing errors or sampling bias. Here we chose either the most commonly estimated frequency for each population with multiple estimates, or the data based on the largest sample size. For instance, the frequencies are all around 17% in 11 Korean population samples; therefore, two estimates of 3% and 36% were rejected in constructing our map. Among nine Japanese samples from Tokyo, with the frequency ranging from 21.5% to 29.0%, we only chose the frequency of 26.6% with the largest sample size of 642. In some cases, we prefer random sample data to the control sample data of case-control studies, or recent data by new typing methods to the data published decades ago. All of the data are included in Table S1 with indication of which samples were included in Figure 1.
