Advancing computational biology and bioinformatics research through open innovation competitions
Fig 3
Gene inference challenge outcomes.
(A): Scatter plot showing the average performance of the top 5 submissions, the benchmark, and an ensemble that takes the best-predicting algorithm from the top 5, where the performance is represented in two dimensions: absolute accuracy (gene-level Spearman correlation) and relative accuracy (gene-level relative rank); curves (light gray) represent locus of points on the plane giving the same aggregate score (curves on the right of this plot represent higher aggregate scores); (B): Boxplot showing absolute accuracy at the gene level for all the top 5 submissions, the benchmark, and the ensemble; (C): Boxplot showing relative accuracy at the gene level for all the top 5 submissions, the benchmark, and the ensemble; and (D): Recall (defined as proportion of genes with a relative accuracy below .95) for all the top 5 submissions, the benchmark, and the ensemble.