Identification and analysis of types of biological protein-protein interactions and their interfaces to predict obligate and non-obligate complexes is a problem that has drawn the attention of the research community in the past few years. In this paper, we propose a prediction approach to predict these two types of complexes. We use desolvation energies - amino acid and atom type - of the residues present in the interface. The prediction is performed via two state-of-the-art classification techniques, namely linear dimensionality reduction (LDR) and support vector machines (SVM). The results on a newly compiled data set, namely BPPI, which is a joint and modified version of two well-known data sets consisting of 213 obligate and 303 non-obligate complexes, show that the best prediction is achieved with SVM (76.94% accuracy) when using desolvation energies of atom-type features. Also, the proposed approach outperforms the previous solvent accessible area-based approaches using SVM (75% accuracy) and LDR (73.06% accuracy). Moreover, a visual analysis of desolvation energies in obligate and non-obligate complexes shows that a few atom-type pairs are good descriptors for these types of complexes.

No abstract

An important issue in understanding and classifying proteinprotein interactions (PPI) is to characterize their interfaces in order to discriminate between transient and obligate complexes. We propose a classification approach to discriminate between these two types of complexes. Our approach uses contact and binding free energies of the residues present in the interaction, which are the input features for the classifiers. A total of 282 features are extracted for each complex, and the classification is performed via recently proposed dimensionality reduction (LDR) methods, including the well-know Fisher's discriminant analysis and two heteroscedastic approaches. The results on a standard benchmark of transient and obligate protein complexes show that LDR approaches achieve a very high classification accuracy (over 78%), outperforming various support vector machines and nearest-neighbor classifiers. An additional insight on the proposed approach and experiments on different subsets of features shows that solvation energies can be used in the classification, leading to a performance comparable to using the full binding free energies of the interaction.

Prediction of protein-protein interactions are important to understand any biological processes. The structural models of the complexes resulting from these interactions are necessary to understand those processes at the molecular level. Xray crystallography is the most popular method to determine the three dimensional structures of protein complexes. However, some of the observed interactions in the structures of protein complexes determined by X-ray crystallography are crystal packing contacts and are not biologically relevant. Thus, it is important to discriminate between biologically relevant interactions and crystal packing contacts. We propose a classification approach to predict these two types of complexes. Our approach has two main features. Firstly, we have calculated various interface property features from the quaternary structures of these interactions. Various features are extracted for each complex, namely numberbased and area-based amino acid compositions. Secondly, these features are treated as the input features of the classifiers. The classification is performed with support vector machines (SVM) and linear dimensionality reduction (LDR) coupled with Bayesian classifiers. The results on a standard benchmark dataset of crystal packing and biological protein complexes show increasing prediction accuracy when compared.