ipPCA analysis of simulated data model 2 with 5 subpopulations. A) Consensus subpopulation tree on ten replicated ipPCA runs. Each cell contains labels SP-1, SP-2, SP-3, SP-4 or SP-5, which refer to subpopulation labels used in simulation. The number of PCs used for clustering is indicated in parentheses in each cell. The blue cell indicates the entire dataset. Nested datasets containing unresolved structure are in green, while the terminated red cells represent resolved subpopulations. B) Scatter-plot using the first and second principal components (PC1 vs. PC2) of the entire dataset (iteration 0 of ipPCA). Each datum point represents an individual. Each subpopulation label is denoted by a separate symbol (see inset). The variation captured by each PC is indicated in parenthesis next to the axis label.