Jewish population structure. (a) Inference using unsupervised clustering. (b) Multidimensional scaling analysis for Jewish individuals. In the unsupervised clustering analysis, the predefined number of clusters was K = 2. The most frequent mode found by Structure for K = 2 included all 40 replicates. For higher values of K, only modes with small numbers of replicates were found (not shown).
Kopelman et al. BMC Genetics 2009 10:80 doi:10.1186/1471-2156-10-80