Fig. 4
From: Benchmark of tools for in silico prediction of MHC class I and class II genotypes from NGS data
Accuracies of meta-prediction models with an increasing number of included tools. Tools were added one by one to the consensus metaclassifier model. At each step, the prediction accuracies of the best performing metaclassifier model for a given number of tools were plotted at the top of the figure. Unfilled markers are placed at the smallest number of tools where the maximal accuracy was obtained for that gene. Black lines indicate the average accuracy of the consensus predictions for the two MHC classes (averaged over all genes of that class). The table below the plot indicates which tools were selected in each model for a given number of tools