Fig. 5
From: A simplicial complex-based approach to unmixing tumor progression data
Comparative accuracy of inferring subpopulation genomic signatures (cell types) from simulated data as a function of data noise level. Accuracy is quantified by the RMSD per model component per dimension, shown in the example cases of two lines at a point (a), two tetrahedra at a point (b), two triangles at a point (c), and two triangles at an edge (d). The inferences are computed as the mean over 10 replicates. The 10 replicates are used to form the merged consensus simplicial complex model. The figure shows relative inference accuracies for the simplicial complex method and the comparative simplex and Gaussian mixture model methods