Figure 3

Evaluation of sequence-based methods in predicting HIV-1 intra-protein coevolution. Distribution plots of Euclidean distance between position pairs in the L top-ranked couplings predicted by individual methods. X- and y-axes indicate the estimated Euclidean distances and the percentage of top-ranked couplings, respectively. Black lines indicate the mean values of Euclidean distances calculated using the L top-ranked couplings. For any method, a lower value of average Euclidean distance indicates that predicted coevolving pairs are in proximity, showing a better prediction performance.