APIS: accurate prediction of hot spots in protein interfaces by combining protrusion index with solvent accessibility

Abstract

BackgroundIt is well known that most of the binding free energy of protein interaction is contributed by a few key hot spot residues. These residues are crucial for understanding the function of proteins and studying their interactions. Experimental hot spots detection methods such as alanine scanning mutagenesis are not applicable on a large scale since they are time consuming and expensive. Therefore, reliable and efficient computational methods for identifying hot spots are greatly desired and urgently required.ResultsIn this work, we introduce an efficient approach that uses support vector machine (SVM) to predict hot spot residues in protein interfaces. We systematically investigate a wide variety of 62 features from a combination of protein sequence and structure information. Then, to remove redundant and irrelevant features and improve the prediction performance, feature selection is employed using the F-score method. Based on the selected features, nine individual-feature based predictors are developed to identify hot spots using SVMs. Furthermore, a new ensemble classifier, namely APIS (A combined model based on Protrusion Index and Solvent accessibility), is developed to further improve the prediction accuracy. The results on two benchmark datasets, ASEdb and BID, show that this proposed method yields significantly better prediction accuracy than those previously published in the literature. In addition, we also demonstrate the predictive power of our proposed method by modelling two protein complexes: the calmodulin/myosin light chain kinase complex and the heat shock locus gene products U and V complex, which indicate that our method can identify more hot spots in these two complexes compared with other state-of-the-art methods.ConclusionWe have developed an accurate prediction model for hot spot residues, given the structure of a protein complex. A major contribution of this study is to propose several new features based on the protrusion index of amino acid residues, which has been shown to significantly improve the prediction performance of hot spots. Moreover, we identify a compact and useful feature subset that has an important implication for identifying hot spot residues. Our results indicate that these features are more effective than the conventional evolutionary conservation, pairwise residue potentials and other traditional features considered previously, and that the combination of our and traditional features may support the creation of a discriminative feature set for efficient prediction of hot spot residues. The data and source code are available on web site http://home.ustc.edu.cn/~jfxia/hotspot.html.

Semi-Supervised Transductive Hot Spot Predictor Working on Multiple Assumptions

“…This is another piece of strong evidence that by employing the three semi-supervised assumptions simultaneously, strong improvements can be achieved. [15], and APIS [16]. Most of these methods are based on machine learning models that encode structural information.…”

“…The existing methods can mainly be classified into three categories, i.e., energy-based methods [4][5][6][7], molecular dynamics-based methods [8][9][10], and machine learning methods [3,[11][12][13][14][15][16][17]. Although energy-based methods and molecular dynamics-based methods demonstrated their power on some tasks, machine learning methods appeared to have better accuracy and scalability.…”

“…In [18], Li et al proposed the geometrically centered region (GCR) based on tripartite graphs, by exploring the role of immobilized water and capturing the compactness in contact and the far distance from bulk solvent, for hot spot prediction. In [16], Xia et al introduced an efficient approach that used SVMs to predict hot spot residues in protein interfaces. In [17], Tuncbag et al presented a web server, HotPoint, which predicted hot spots in protein interfaces using an empirical model.…”

Computational Tools and Databases for the Study and Characterization of Protein Interactions

“…These methodologies range from scoring function derived from simple physical models (Guerois et al 2002;Kortemme and Baker 2002;Kruger and Gohlke 2010) to more complex, time consuming atomistic simulations to model effect of mutations in the binding energy (Almlof et al 2006;Lafont et al 2007;Moreira et al 2007;Benedix et al 2009;Diller et al 2010). Other methods exploit individual features (or combination of them) that are characteristic to hot spots such as solvent accessibility (Landon et al 2007;Tuncbag et al 2009;Xia et al 2010;, atomic contacts , structural conservation (Li et al 2004), restricted mobility (Yogurtcu et al 2008), relative location of residues in the interface (Keskin et al 2005), sequence conservation (Hu et al 2000;Ma and Nussinov 2007) and pattern mining (Hsu et al 2007). Other examples include a number of machine learning approaches (Darnell et al 2007;Cho et al 2009;Lise et al 2009;Assi et al 2010) such as PCRPi (see next) that integrate a range of structural-and sequence-based information and a docking-based approach (Grosdidier and Fernandez-Recio 2008).…”

Boosting Prediction Performance of Protein–Protein Interaction Hot Spots by Using Structural Neighborhood Properties

“…On the other hand, empirical functions or simple physical methods, such as FOLDEF (Guerois et al, 2002) and Robetta (Kortemme and Baker, 2002), which use experimentally calibrated knowledge-based simplified models to evaluate the binding free energy, provide an alternative way to probe hot spots with much less computation. Recently, there has been considerable interest applying machine-learning methods to predict hot spots such as neural networks (Ofran and Rost, 2007), decision trees (Darnell et al, 2007), support vector machines (Cho et al, 2009;Xia et al, 2010;Zhu and Mitchell, 2011), Bayesian networks (Assi et al, 2009), minimum cut trees (Tuncbag et al, 2010), and random forests (Wang et al, 2012).…”