Advancing computational biology and bioinformatics research through open innovation competitions
Fig 4
Gene inference challenge: Solutions clustering.
(A): Plot showing results of the Principal Component Analysis (PCA) of the gene-level combined scores of all top five submissions for the CMap Inference Challenge. While all top submissions are clustered together in the first factor of PCA, they differ in the second factor, separating into different “clusters” the Neural Network approach (submission 2) and the other KNN based approaches (submissions 1, 3, 4, and 5). (B): Barplot showing the proportion (and count in parenthesis) of gene predictions used to form the ensemble, which uses the training dataset to select the best predicting algorithm from the MLR benchmark and the top 5 submissions (i.e., the algorithm achieving the highest combined score for a particular gene) for the CMap Inference Challenge.