Mina Maleki

Prediction of biological protein–protein interactions using atom-type and amino acid properties

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.

The role of electrostatic energy in prediction of obligate protein-protein interactions.

BackgroundPrediction and analysis of protein-protein interactions (PPI) and specifically types of PPIs is an important problem in life science research because of the fundamental roles of PPIs in many biological processes in living cells. In addition, electrostatic interactions are important in understanding inter-molecular interactions, since they are long-range, and because of their influence in charged molecules. This is the main motivation for using electrostatic energy for prediction of PPI types.ResultsWe propose a prediction model to analyze protein interaction types, namely obligate and non-obligate, using electrostatic energy values as properties. The prediction approach uses electrostatic energy values for pairs of atoms and amino acids present in interfaces where the interaction occurs. The main features of the complexes are found and then the prediction is performed via several state-of-the-art classification techniques, including linear dimensionality reduction (LDR), support vector machine (SVM), naive Bayes (NB) and k-nearest neighbor (k-NN). For an in-depth analysis of classification results, some other experiments were performed by varying the distance cutoffs between atom pairs of interacting chains, ranging from 5Å to 13Å. Moreover, several feature selection algorithms including gain ratio (GR), information gain (IG), chi-square (Chi2) and minimum redundancy maximum relevance (mRMR) are applied on the available datasets to obtain more discriminative pairs of atom types and amino acid types as features for prediction.ConclusionsOur results on two well-known datasets of obligate and non-obligate complexes confirm that electrostatic energy is an important property to predict obligate and non-obligate protein interaction types on the basis of all the experimental results, achieving accuracies of over 98%. Furthermore, a comparison performed by changing the distance cutoff demonstrates that the best values for prediction of PPI types using electrostatic energy range from 9Å to 12Å, which show that electrostatic interactions are long-range and cover a broader area in the interface. In addition, the results on using feature selection before prediction confirm that (a) a few pairs of atoms and amino acids are appropriate for prediction, and (b) prediction performance can be improved by eliminating irrelevant and noisy features and selecting the most discriminative ones.

Analysis of relevant physicochemical properties in obligate and non-obligate protein-protein interactions

Identification and analysis of types of protein-protein interactions (PPI) is an important problem in molecular biology because of its key role in many biological processes in living cells. In this paper, we focus on obligate and non-obligate complexes, their prediction and analysis. We propose a feature selection scheme called MRMRpro which is based on Minimum Redundancy Maximum Relevance (MRMR) to focus on the most discriminative and relevant properties to distinguish between these two types of complexes. Our prediction approach uses desolvation energies of pairs of atoms or amino acids present in the interfaces of such complexes. Our results on two well-known datasets confirm that MRMRpro leads to significant improvements on performance by finding more relevant features for prediction. Furthermore, the prediction performance of our biologically guided feature selection methods demonstrate that hydrophobic amino acids are more discriminating than hydrophilic and amphipathic amino acids to distinguish between obligate and non-obligate complexes.

Using desolvation energies of structural domains to predict stability of protein complexes

Employing domain knowledge for prediction of particular types of protein–protein interactions (PPIs) is a problem that has become increasingly important in the past few years, due to the fundamental role of domains in protein function. We propose a model to predict obligate and non-obligate protein interaction types using desolvation energies of structural domains that are present in the interfaces of protein complexes, which are extracted from the CATH database. The prediction is performed using several state-of-the-art classification techniques, including linear dimensionality reduction, a support vector machine based on sequential minimal optimization, naive Bayes, and k-nearest neighbour. Our results on two well-known datasets demonstrate that (a) domain-based features of higher levels of CATH, especially level 2, are more powerful and discriminative than features of other levels, and (b) properties taken from different levels of the CATH hierarchy yield higher accuracies than properties taken from each level of the hierarchy separately. Furthermore, analysis of structural properties suggests that domain–domain interactions that have at least a mainly-beta secondary structure in one sub-unit are more informative for predicting obligate and non-obligate PPIs.

Multi-level structural domain-domain interactions for prediction of obligate and non-obligate protein-protein interactions

The prediction of particular types of protein-protein interactions (PPIs) based on knowledge of their interacting domains is a problem that has drawn the attention of researchers in the past few years. We focus on the prediction and analysis of obligate and nonobligate complexes by using structural domains from the CATH database. Our proposed prediction model is an extension of the one used in our previous work, which uses desolvation energies of domain-domain interactions (DDIs) present in the interfaces of such complexes. The prediction is performed via a support vector machine (SVM). Whereas previous efforts have considered structural domains taken from each level of the CATH hierarchy, in turn, we generalize this to allow some domains to be considered at one level of the hierarchy, while others may be considered at a different level. Our results show an improvement of 4.58% and 2.36% on two well-known datasets over our previous results.

Analysis of obligate and non-obligate complexes using desolvation energies in domain-domain interactions

Protein-protein interactions (PPI) are important in most biological processes and their study is crucial in many applications. Identification of types of protein complexes is a particular problem that has drawn the attention of the research community in the past few years. We focus on obligate and non-obligate complexes, their prediction and analysis. We propose a prediction model to distinguish between these two types of complexes, which uses desolvation energies of domain-domain interactions (DDI), pairs of atoms and amino acids present in the interfaces of such complexes. Principal components of the data were found and then the prediction is performed via linear dimensionality reduction (LDR) and support vector machines (SVM). Our results on a newly compiled dataset, namely binary-PPID, which is a joint and modified version of two well-known datasets consisting of 146 obligate and 169 non-obligate complexes, show that the best prediction is achieved with SVM (77.78%) when using desolvation energies of atom type features. Furthermore, a detailed analysis shows that different DDIs are present in obligate and non-obligate complexes, and that homo-DDIs are more likely to be present in obligate interactions.

Domain-domain interactions in obligate and non-obligate protein-protein interactions

In this study, an analysis of protein-protein interactions (PPIs) that uses properties of domain-domain interactions (DDIs) present in the interface is discussed. The aim is to predict obligate and non-obligate complexes. The results show that support vector machines (SVM) classifier achieves much better prediction performance, even better than linear dimensionality reduction (LDR) schemes and aslo desolvation energy is better than interface area and composition for predicting transient and obligate complexes. Moreover, a visual and numerical analysis insight of the distribution of the DDIs in different complexes is shown that most homo-domain pairs are in obligate interactions.

Computational analysis of the stability of SCF ligases employing domain information

Because of the unequivocally fundamental role of SCF ubiquitin ligase in many biological functions within a living cell including regulating DNA repair, cell cycle progression, and inflammation, we have analyzed the role of domain interactions in determining particular types of protein-protein interactions (PPIs) that are known or predicted to occur involving subunit components of the SCF-ligase complex. We focus on the prediction and analysis of obligate and non-obligate SCF-ligase complexes by using sequence domains from the Pfam database. After extracting different types of feature vectors, the prediction is performed via a support vector machine (SVM). The numerical results demonstrate that most of the interactions of SCF-ligase complexes are mediated by at least one domain. Moreover, domain-domain interactions dominate in obligate complexes whereas non-obligate complexes exhibit more domain-peptide chain interactions. Also, the computational results show that the best prediction accuracy of 80.46% is achieved using the combination of feature vectors of domain-domain type, domain-peptide chain type and no-domain interactions.

Using structural domains to predict obligate and non-obligate protein-protein interactions

The identification and prediction of particular types of protein-protein interactions (PPIs) based on knowledge of their interacting domains is a problem that has drawn the attention of researchers in the past few years. We focus on the prediction and analysis of obligate and non-obligate complexes by using structural domains from the CATH database. Our proposed prediction model uses desolvation energies of domain-domain interactions (DDIs) present in the interfaces of such complexes. The prediction is performed via linear dimensionality reduction (LDR) and support vector machines (SVMs). Our results on two well-known datasets show that DDI features of the first three levels of CATH, especially level 2, are more powerful and discriminative than features of other levels in predicting these types of complexes. Furthermore, a detailed analysis shows that different DDIs are present in obligate and non-obligate complexes, and that homo-DDIs are more likely to be present in obligate interactions.

A Computational Domain-Based Feature Grouping Approach for Prediction of Stability of SCF Ligases

Analyzing the stability of SCF ubiquitin ligases is worth investigating because these complexes are involved in many cellular processes including cell cycle regulation, DNA repair mechanisms, and gene expression. On the other hand, interactions of two (or more) proteins are controlled by their domains – compact functional units of proteins. As a consequence, in this study, we have analyzed the role of Pfam domain interactions in predicting the stability of protein-protein interactions (PPIs) that are known or predicted to occur involving subunit components of the SCF ligase complex. Moreover, employing the most relevant and discriminating features is very important to achieve a successful prediction with low computational cost. Although, different feature selection methods have been recently developed for this purpose, feature grouping is a better idea, especially when dealing with high-dimensional sparse feature vectors, yielding better interpretation of the data. In this paper, a correlation-based feature grouping (CFG) method is proposed to group and combine the features. To demonstrate the strength of CFG, two filter methods of χ 2 and correlation are also employed for feature selection and prediction is performed using different methods including a support vector machine (SVM) and k-Nearest Neighbor (k-NN). The experimental results on a dataset of SCF ligases indicate that employing feature grouping achieves significant increases of 10% for svm and 13% for k-NN, being more efficient than employing feature selection in identifying a set of relevant features