Additional file 1: of Genetic differences among ethnic groups

The predictive performance of the SMO method on 30 randomly divided training and independent test datasets. The training (85 % of the samples) and independent test (15 % of the samples) datasets were randomly divided for 30 times and for each time, the training and test processes were repeated. Then the mean and standard deviation of the accuracies on 30 training and independent test datasets were calculated. As shown in this table, the mean accuracies were close to the accuracies of SMO in Table 2 and the standard deviations were very small which indicated that the partition of training and independent test datasets does not affect the prediction performance. (XLSX 20 kb)