smoothing approach, our method only uses a subset of autosome markers that are at least 1 Mb away from each other when building the regression model. While this simple approach circumvents the problem of dependency between neighboring markers, this also raises a concern on the model stability. To investigate this issue, we tested building models with 10 different sets of markers on the two samples used in Figure 2, and found that the regression model parameters are quite stable and are little affected by the use of different markers in the training process. The resulting GCWF values after adjustment are virtually identical.
