Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: Decoy selection for protein structure prediction via extreme gradient boosting and ranking

Fig. 4

Comparison of the five selection strategies ML-Select, Size (S), Size+Energy (S+E), Pareto-Rank (PR), and Pareto-Rank+Count (PR+PC), in terms of the p metric, for the easy, medium, and hard test cases. The top row shows the results for easy cases, second row is for the medium cases, and the bottom row shows the results for the hard cases. Metric p, purity, measures the percentage of near-native decoys in the x selected basins while penalizing the basins by the extent of false positive presence. Results are shown for x∈{1,3}

Back to article page