Skip to main content
Figure 2 | BMC Bioinformatics

Figure 2

From: Correlated mutations via regularized multinomial regression

Figure 2

Performance of RMRCM compared to performance of MI on artificial datasets. (A) Representative example of ROC curves for RMRCM (dashed) and MI (continuous line) for simulated dataset with network edge density 0.25, and 500 sequences in the multiple sequence alignment. Performance is assessed by comparing predicted contacts with those in the network used to generate the artificial sequences. (B) AUC values for MI (x-axis) vs. RMRCM (y-axis) for datasets with 500 or 1,000 sequences. Arrow indicates particular case illustrated in panel A. Black indicates data without noise, red with "position-noise" (10% and 25%) for interaction density 0.25 and blue with "sequence-noise" (10% and 25%) for interaction density 0.1.

Back to article page