The effect of ethnic and racial population substructuring on the estimation of multi-locus fixed-bin VNTR RFLP genotype probabilities.

Four East Asian ethnic and four racial VNTR RFLP Southern California databases were used to determine the impact of population substructure on fixed-bin genotype probability estimates. Two calculations were used for population-level probabilities: Stratified sampling, which takes substructuring into account, and pooling, which ignores it. Using 1000 four-locus genotypes, the relative difference between probabilities calculated with the stratified and the pooled methods did not exceed one order of magnitude out of about 11 orders of magnitude for East-Asian racial genotypes. Pooled estimates differed from cognate ethnic values by less than one order of magnitude out of about six. These findings suggest substructuring of races by major ethnic groups does not lead to large errors. Racial genotype probability variances were on average about twice the ethnic variances. Multi-racial total population probabilities calculated by the pooled and stratified methods differed by less than one order of magnitude out of five.

Diversity of ethnic and racial VNTR RFLP fixed-bin frequency distributions.

To examine the impact that intra- and interracial genetic diversities have on VNTR RFLP-fragment-size distributions, a multiracial (East Asian, African American, U.S. Southwest Hispanic, and European Caucasian) and multiethnic (Chinese, Japanese, Korean, and Vietnamese) database has been constructed for the following loci: D1S7, D2S44, D4S139, and D10S28. Homogeneity between samples was examined using the Komologorov-Smirnov two-sample test for RFLP fragment sizes and a log-likelihood test for fixed-bin frequencies with theoretical and Monte Carlo empirical significance levels. Small but significant differences between theoretical and empirical significance-level distributions were observed with both procedures, particularly with the latter. The significance levels of the two types of tests were poorly correlated. Statistically significant differences in fragment-size and fixed-bin distributions were found within and between races, with greater differences occurring between races. Cluster analysis and principal components analysis, using different similarity measures, did not support the hypothesis of greater intra- than interracial diversity, which suggests that ethnic variation can be conservatively estimated by racial variation.