Figure 9
From: Iterative pruning PCA improves resolution of highly structured populations
Population assignment comparison of Shriver's dataset among ipPCA, STRUCTURE, BAPS, and non-iterative PCA clustering algorithms. A) Log probability data plot of inferred K for STRUCTURE results (no admixture model) for K = 8 to K = 20; STRUCTURE parameters: 100000 simulation iterations for burn-ins and 100000 additional iterations for parameter estimation. B) Side-by-side comparison of subpopulation assignment by ipPCA, STRUCTURE, BAPS and non-iterative PCA clustering (NI-PCA). Population labels are the same as in [36].