Fig. 7
From: CoINcIDE: A framework for discovery of patient subtypes across multiple datasets
CoINcIDE ovarian short and long gene list meta-clusters. CoINcIDE results on 24 ovarian datasets from de novo clusterings using k-means consensus with the PAC score. a, c Resulting cluster-cluster networks for the short and long gene list clusterings, respectively. Node size is scaled to the relative number of samples in each cluster. b, d Five-year survival curves for the short and long gene list clusterings, respectively, as stratified by the final CoINcIDE subtypes depicted in (a) and (c). The symbols along the survival curves represent actual patient dataset points. A minimum mean similarity metric threshold of 0.5 was used to derive the long gene list results (Additional file 3: Figures S17 and S18 show results using a threshold of 0.7). N = number of events used in the model