Supplementary Figure 1: ParsSNP convergence and reproducibility.
From: Unsupervised detection of cancer driver mutations with parsimony-guided learning
(a) The EM portion of ParsSNP consistently converges in 15–20 iterations. Lines are offset slightly to aid visualization. (b) The pan-cancer training set was partitioned randomly into two equally sized, independent halves. ParsSNP produces highly correlated scores when trained on independent but comparable data sets (n = 566,223).
