Sample predictions. One protein is selected from each complexes and shown in sub-figure (A), (B), (C) and (D). The PDB IDs and chain identifiers are 1bplB, 1ijeB, 1lqpB and 1j0xO respectively. The interface residues are depicted with purple colour. For each sub-figure, the true interfaces are shown in the center picture. The left picture gives the results predicted by the second SVM, which takes residue interface propensities as an extra feature. The right picture gives the results predicted by the third SVM, which takes binary profile propensities as an extra feature.