Predicting Protein Interaction Sites Research Articles

Predicting Protein Interaction Sites Using PITHIA.

Several proteins work independently, but the majority work together to maintain the functions of the cell. Thus, it is crucial to know the interaction sites that facilitate protein-protein interactions. The development of effective computational methods is essential because experimental methods are expensive and time-consuming. This chapter is a guide to predicting protein interaction sites using the program "PITHIA." First, some installation guides are presented, followed by descriptions of input file formats. Afterward, PITHIA's commands and options are outlined with examples. Moreover, some notes are provided on how to extend PITHIA's installation and usage.

Protein-Protein Interaction Sites Prediction Based on an Under-Sampling Strategy and Random Forest Algorithm.

The computational methods of protein-protein interaction sites prediction can effectively avoid the shortcomings of high cost and time in traditional experimental approaches. However, the serious class imbalance between interface and non-interface residues on the protein sequences limits the prediction performance of these methods. This work therefore proposed a new strategy, NearMiss-based under-sampling for unbalancing datasets and Random Forest classification (NM-RF), to predict protein interaction sites. Herein, the residues on protein sequences were represented by the PSSM-derived features, hydropathy index (HI) and relative solvent accessibility (RSA). In order to resolve the class imbalance problem, an under-sampling method based on NearMiss algorithm is adopted to remove some non-interface residues, and then the random forest algorithm is used to perform binary classification on the balanced feature datasets. Experiments show that the accuracy of NM-RF model reaches 87.6% and 84.3% on Dtestset72 and PDBtestset164 respectively, which demonstrate the effectiveness of the proposed NM-RF method in differentiating the interface or non-interface residues.

Predicting protein–protein interaction sites using modified support vector machine

Protein–protein interaction plays a fundamental role in many biological processes and diseases. Characterizing protein interaction sites is crucial for the understanding of the mechanism of protein–protein interaction and their cellular functions. In this paper, we proposed a method based on integrated support vector machine (SVM) with a hybrid kernel to predict protein interaction sites. First, a number of features of the protein interaction sites were extracted. Secondly, the technique of sliding window was used to construct a protein feature space based on the influence of the adjacent residues. Thirdly, to avoid the impact of imbalance of the data set on prediction accuracy, we employed boost-strap to re-sample the data. Finally, we built a SVM classifier, whose hybrid kernel comprised a Gaussian kernel and a Polynomial kernel. In addition, an improved particle swarm optimization (PSO) algorithm was applied to optimize the SVM parameters. Experimental results show that the PSO-optimized SVM classifier outperforms existing methods.

DBSI: DNA-binding site identifier

In this study, we present the DNA-Binding Site Identifier (DBSI), a new structure-based method for predicting protein interaction sites for DNA binding. DBSI was trained and validated on a data set of 263 proteins (TRAIN-263), tested on an independent set of protein-DNA complexes (TEST-206) and data sets of 29 unbound (APO-29) and 30 bound (HOLO-30) protein structures distinct from the training data. We computed 480 candidate features for identifying protein residues that bind DNA, including new features that capture the electrostatic microenvironment within shells near the protein surface. Our iterative feature selection process identified features important in other models, as well as features unique to the DBSI model, such as a banded electrostatic feature with spatial separation comparable with the canonical width of the DNA minor groove. Validations and comparisons with established methods using a range of performance metrics clearly demonstrate the predictive advantage of DBSI, and its comparable performance on unbound (APO-29) and bound (HOLO-30) conformations demonstrates robustness to binding-induced protein conformational changes. Finally, we offer our feature data table to others for integration into their own models or for testing improved feature selection and model training strategies based on DBSI.

Current Status of Machine Learning-Based Methods for Identifying Protein-Protein Interaction Sites

High-throughput experimental technologies continue to alter the study of current system biology. Investigators are understandably eager to harness the power of these new technologies. Protein-protein interactions on these platforms, however, present numerous production and bioinformatics challenges. Some issues like feature extraction, feature representation, prediction algorithm and results analysis have become increasingly problematic in the prediction of protein-protein interaction sites. The development of powerful, efficient prediction methods for inferring protein interface residues based on protein primary sequence or/and 3D structure is critical for the research community to accelerate research and publications. Currently, machine learning-based approaches are drawing the most attention in predicting protein interaction sites. This review aims to describe the state of the whole pipeline when machine learning strategies are applied to infer protein interaction sites.

MetaPIS: A Sequence-based Meta-server for Protein Interaction Site Prediction

The identification of interfaces in protein complexes is effective for the elucidation of protein function and helps us to understand their roles in biological processes. With the exponentially growing amount of protein sequence data, an exploration of new methods that predict protein interaction sites based solely on sequence information is becoming increasingly urgent. Because a combination of different methods could produce better results than a single method, interaction site prediction can be improved through the utilization of different methods. This paper describes a new method that predicts interaction sites based on protein sequences by integrating five different algorithms employing meta-method, Majority Vote and SVMhmm Regression techniques. The 'metaPIS' web-server was implemented for meta-prediction. An evaluation of the meta-methods using independent datasets revealed that Majority Vote achieved the highest average Matthews correlation coefficient (0.181) among all the methods assessed. SVMhmm Regression achieved a lower score but provided a more stable result. The metaPIS server allows experimental biologists to speculate regarding protein function by identifying potential interaction sites based on protein sequence. As a web server, metaPIS is freely accessible to the public at http://202.116.74.5:84/metapis.

MetaPIS: A Sequence-based Meta-server for Protein Interaction Site Prediction

The identification of interfaces in protein complexes is effective for the elucidation of protein function and helps us to understand their roles in biological processes. With the exponentially growing amount of protein sequence data, an exploration of new methods that predict protein interaction sites based solely on sequence information is becoming increasingly urgent. Because a combination of different methods could produce better results than a single method, interaction site prediction can be improved through the utilization of different methods. This paper describes a new method that predicts interaction sites based on protein sequences by integrating five different algorithms employing meta-method, Majority Vote and SVMhmm Regression techniques. The 'metaPIS' web-server was implemented for meta-prediction. An evaluation of the meta-methods using independent datasets revealed that Majority Vote achieved the highest average Matthews correlation coefficient (0.181) among all the methods assessed. SVMhmm Regression achieved a lower score but provided a more stable result. The metaPIS server allows experimental biologists to speculate regarding protein function by identifying potential interaction sites based on protein sequence. As a web server, metaPIS is freely accessible to the public at http://202.116.74.5:84/metapis. Keywords: Classification method, interaction site prediction, machine learning, meta-method, protein sequence, sequential labeling method, web server

Predicting Protein Interaction Sites Based on a New Integrated Radial Basis Functional Neural Network

Interactions among proteins are the basis of various life events. So, it is important to recognize and research protein interaction sites. A control set that contains 149 protein molecules were used here. Then 10 features were extracted and 4 sample sets that contained 9 sliding windows were made according to features. These 4 sample sets were calculated by Radial Basis Functional neutral networks which were optimized by Particle Swarm Optimization respectively. Then 4 groups of results were obtained. Finally, these 4 groups of results were integrated by Genetic Algorithm based Selected Ensemble (GASEN) and better accuracy was got. So, the integrated method was proved to be effective.

Predicting Protein Interaction Sites Based on a New Integrated Radial Basis Functional Neural Network

Interactions among proteins are the basis of various life events. So, it is important to recognize and research protein interaction sites. A control set that contains 149 protein molecules were used here. Then 10 features were extracted and 4 sample sets that contained 9 sliding windows were made according to features. These 4 sample sets were calculated by Radial Basis Functional neutral networks which were optimized by Particle Swarm Optimization respectively. Then 4 groups of results were obtained. Finally, these 4 groups of results were integrated by Genetic Algorithm based Selected Ensemble (GASEN) and better accuracy was got. So, the integrated method was proved to be effective.

Sequence-based identification of interface residues by an integrative profile combining hydrophobic and evolutionary information

BackgroundProtein-protein interactions play essential roles in protein function determination and drug design. Numerous methods have been proposed to recognize their interaction sites, however, only a small proportion of protein complexes have been successfully resolved due to the high cost. Therefore, it is important to improve the performance for predicting protein interaction sites based on primary sequence alone.ResultsWe propose a new idea to construct an integrative profile for each residue in a protein by combining its hydrophobic and evolutionary information. A support vector machine (SVM) ensemble is then developed, where SVMs train on different pairs of positive (interface sites) and negative (non-interface sites) subsets. The subsets having roughly the same sizes are grouped in the order of accessible surface area change before and after complexation. A self-organizing map (SOM) technique is applied to group similar input vectors to make more accurate the identification of interface residues. An ensemble of ten-SVMs achieves an MCC improvement by around 8% and F1 improvement by around 9% over that of three-SVMs. As expected, SVM ensembles constantly perform better than individual SVMs. In addition, the model by the integrative profiles outperforms that based on the sequence profile or the hydropathy scale alone. As our method uses a small number of features to encode the input vectors, our model is simpler, faster and more accurate than the existing methods.ConclusionsThe integrative profile by combining hydrophobic and evolutionary information contributes most to the protein-protein interaction prediction. Results show that evolutionary context of residue with respect to hydrophobicity makes better the identification of protein interface residues. In addition, the ensemble of SVM classifiers improves the prediction performance.AvailabilityDatasets and software are available at http://mail.ustc.edu.cn/~bigeagle/BMCBioinfo2010/index.htm.

Prediction of Protein-Protein Interaction Sites Using Electrostatic Desolvation Profiles

Protein-protein complex formation involves removal of water from the interface region. Surface regions with a small free energy penalty for water removal or desolvation may correspond to preferred interaction sites. A method to calculate the electrostatic free energy of placing a neutral low-dielectric probe at various protein surface positions has been designed and applied to characterize putative interaction sites. Based on solutions of the finite-difference Poisson equation, this method also includes long-range electrostatic contributions and the protein solvent boundary shape in contrast to accessible-surface-area-based solvation energies. Calculations on a large set of proteins indicate that in many cases (>90%), the known binding site overlaps with one of the six regions of lowest electrostatic desolvation penalty (overlap with the lowest desolvation region for 48% of proteins). Since the onset of electrostatic desolvation occurs even before direct protein-protein contact formation, it may help guide proteins toward the binding region in the final stage of complex formation. It is interesting that the probe desolvation properties associated with residue types were found to depend to some degree on whether the residue was outside of or part of a binding site. The probe desolvation penalty was on average smaller if the residue was part of a binding site compared to other surface locations. Applications to several antigen-antibody complexes demonstrated that the approach might be useful not only to predict protein interaction sites in general but to map potential antigenic epitopes on protein surfaces.

Prediction of protein-protein interaction sites using an ensemble method

BackgroundPrediction of protein-protein interaction sites is one of the most challenging and intriguing problems in the field of computational biology. Although much progress has been achieved by using various machine learning methods and a variety of available features, the problem is still far from being solved.ResultsIn this paper, an ensemble method is proposed, which combines bootstrap resampling technique, SVM-based fusion classifiers and weighted voting strategy, to overcome the imbalanced problem and effectively utilize a wide variety of features. We evaluate the ensemble classifier using a dataset extracted from 99 polypeptide chains with 10-fold cross validation, and get a AUC score of 0.86, with a sensitivity of 0.76 and a specificity of 0.78, which are better than that of the existing methods. To improve the usefulness of the proposed method, two special ensemble classifiers are designed to handle the cases of missing homologues and structural information respectively, and the performance is still encouraging. The robustness of the ensemble method is also evaluated by effectively classifying interaction sites from surface residues as well as from all residues in proteins. Moreover, we demonstrate the applicability of the proposed method to identify interaction sites from the non-structural proteins (NS) of the influenza A virus, which may be utilized as potential drug target sites.ConclusionOur experimental results show that the ensemble classifiers are quite effective in predicting protein interaction sites. The Sub-EnClassifiers with resampling technique can alleviate the imbalanced problem and the combination of Sub-EnClassifiers with a wide variety of feature groups can significantly improve prediction performance.

Prediction of Protein–Protein Interaction Sites in Sequences and 3D Structures by Random Forests

Identifying interaction sites in proteins provides important clues to the function of a protein and is becoming increasingly relevant in topics such as systems biology and drug discovery. Although there are numerous papers on the prediction of interaction sites using information derived from structure, there are only a few case reports on the prediction of interaction residues based solely on protein sequence. Here, a sliding window approach is combined with the Random Forests method to predict protein interaction sites using (i) a combination of sequence- and structure-derived parameters and (ii) sequence information alone. For sequence-based prediction we achieved a precision of 84% with a 26% recall and an F-measure of 40%. When combined with structural information, the prediction performance increases to a precision of 76% and a recall of 38% with an F-measure of 51%. We also present an attempt to rationalize the sliding window size and demonstrate that a nine-residue window is the most suitable for predictor construction. Finally, we demonstrate the applicability of our prediction methods by modeling the Ras–Raf complex using predicted interaction sites as target binding interfaces. Our results suggest that it is possible to predict protein interaction sites with quite a high accuracy using only sequence information.

Prediction of protein-protein binding site by using core interface residue and support vector machine

BackgroundThe prediction of protein-protein binding site can provide structural annotation to the protein interaction data from proteomics studies. This is very important for the biological application of the protein interaction data that is increasing rapidly. Moreover, methods for predicting protein interaction sites can also provide crucial information for improving the speed and accuracy of protein docking methods.ResultsIn this work, we describe a binding site prediction method by designing a new residue neighbour profile and by selecting only the core-interface residues for SVM training. The residue neighbour profile includes both the sequential and the spatial neighbour residues of an interface residue, which is a more complete description of the physical and chemical characteristics surrounding the interface residue. The concept of core interface is applied in selecting the interface residues for training the SVM models, which is shown to result in better discrimination between the core interface and other residues.The best SVM model trained was tested on a test set of 50 randomly selected proteins. The sensitivity, specificity, and MCC for the prediction of the core interface residues were 60.6%, 53.4%, and 0.243, respectively. Our prediction results on this test set were compared with other three binding site prediction methods and found to perform better. Furthermore, our method was tested on the 101 unbound proteins from the protein-protein interaction benchmark v2.0. The sensitivity, specificity, and MCC of this test were 57.5%, 32.5%, and 0.168, respectively.ConclusionBy improving both the descriptions of the interface residues and their surrounding environment and the training strategy, better SVM models were obtained and shown to outperform previous methods. Our tests on the unbound protein structures suggest further improvement is possible.

Predicting protein interaction sites: binding hot-spots in protein–protein and protein–ligand interfaces

Protein assemblies are currently poorly represented in structural databases and their structural elucidation is a key goal in biology. Here we analyse clefts in protein surfaces, likely to correspond to binding 'hot-spots', and rank them according to sequence conservation and simple measures of physical properties including hydrophobicity, desolvation, electrostatic and van der Waals potentials, to predict which are involved in binding in the native complex. The resulting differences between predicting binding-sites at protein-protein and protein-ligand interfaces are striking. There is a high level of prediction accuracy (< or =93%) for protein-ligand interactions, based on the following attributes: van der Waals potential, electrostatic potential, desolvation and surface conservation. Generally, the prediction accuracy for protein-protein interactions is lower, with the exception of enzymes. Our results show that the ease of cleft desolvation is strongly predictive of interfaces and strongly maintained across all classes of protein-binding interface.

Predicting protein interaction sites from residue spatial sequence profile and evolution rate

This paper proposes a novel method that can predict protein interaction sites in heterocomplexes using residue spatial sequence profile and evolution rate approaches. The former represents the information of multiple sequence alignments while the latter corresponds to a residue’s evolutionary conservation score based on a phylogenetic tree. Three predictors using a support vector machines algorithm are constructed to predict whether a surface residue is a part of a protein–protein interface. The efficiency and the effectiveness of our proposed approach is verified by its better prediction performance compared with other models. The study is based on a non-redundant data set of heterodimers consisting of 69 protein chains.

Mapping protein?protein interactions by bioinformatics and cross-linking

Protein–protein interactions play a key role in many cellular processes. Determining the protein interaction network is one of the major tasks of functional proteomics. Databases of protein interactions integrate the large amounts of information available on protein– protein interactions, ranging from functional interactions (such as which proteins are involved in a common metabolic pathway) to physical interactions (like which proteins form a complex) [1]. Whereas the amount of information regarding protein structures and protein– protein interactions is steadily increasing, structural determination of protein–protein complexes using experimental methods such as X-ray crystallography, NMR spectroscopy and electron microscopy is lagging behind [2]. Hence, bioinformatic methods for predicting protein interactions sites and structures of protein complexes are becoming increasingly important. The predictive power of these methods is measured at regular intervals using critical assessment of predicted interactions (CAPRI) [3]. One lesson from these studies is that great benefit can be obtained by combining protein–protein docking with data from other predictive methods and the results of experimental approaches yielding low resolution structural information. Experimental data can be obtained by a variety of methods [2]; here the focus will be on chemical cross-linking, a classical method which has gained renewed attention due to advances in molecular biology, chemical synthesis and mass spectrometric analysis. Bioinformatic methods for predicting protein interaction sites

A fast method to predict protein interaction sites from sequences

A simple method for predicting residues involved in protein interaction sites is proposed. In the absence of any structural report, the procedure identifies linear stretches of sequences as “receptor-binding domains” (RBDs) by analysing hydrophobicity distribution. The sequences of two databases of non-homologous interaction sites eliciting various biological activities were tested; 59–80 % were detected as RBDs. A statistical analysis of amino acid frequencies was carried out in known interaction sites and in predicted RBDs. RBDs were predicted from the 80,000 sequences of the Swissprot database. In both cases, arginine is the most frequently occurring residue. The RBD procedure can also detect residues involved in specific interaction sites such as the DNA-binding (95 % detected) and Ca-binding domains (83 % detected). We report two recent analyses; from the prediction of RBDs in sequences to the experimental demonstration of the functional activities. The examples concern a retroviral Gag protein and a penicillin-binding protein. We support that this method is a quick way to predict protein interaction sites from sequences and is helpful for guiding experiments such as site-specific mutageneses, two-hybrid systems or the synthesis of inhibitors.