On the post-glacial spread of human commensal Arabidopsis thaliana.

We chose K =5 for our analyses because its cross-validation errors: (1) are relative low, (2) do not overlap much with the range of the previous K and (3) have highly significant difference compared with the previous K but not so much with the next K value ( Supplementary Fig. 9a ).

also in 132,142 other papers