Figure 2
From: Mutual information and variants for protein domain-domain contact prediction
Contact versus non-contact prediction P-ROC curves for MI variants on the 40 test cases. A, B and C illustrate the performance of MI, MIp and MIc variants respectively when distinguishing contact from non-contact surface residues. The solid green line in all plots depicts the chance of randomly selecting a contact residue, while the dashed green line indicates the probability of randomly selecting a contact residue when employing the reduced alphabet amino acid set.