Predicting protein function from protein/protein interaction data: a probabilistic approach

MOTIVATION:The development of experimental methods for genome scale analysis of molecular interaction networks has made possible new approaches to inferring protein function. This paper describes a method of assigning functions based on a probabilistic analysis of graph neighborhoods in a protein-protein interaction network. The method exploits the fact that graph neighbors are more likely to share functions than nodes which are not neighbors. A binomial model of local neighbor function labeling probability is combined with a Markov random field propagation algorithm to assign function probabilities for proteins in the network. RESULTS: We applied the method to a protein-protein interaction dataset for the yeast Saccharomyces cerevisiae using the Gene Ontology (GO) terms as function labels. The method reconstructed known GO term assignments with high precision, and produced putative GO assignments to 320 proteins that currently lack GO annotation, which represents about 10% of the unlabeled proteins in S. cerevisiae.     

Prediction of protein function using protein-protein interaction data.

Assigning functions to novel proteins is one of the most important problems in the postgenomic era. Several approaches have been applied to this problem, including the analysis of gene expression patterns, phylogenetic profiles, protein fusions, and protein-protein interactions. In this paper, we develop a novel approach that employs the theory of Markov random fields to infer a protein's functions using protein-protein interaction data and the functional annotations of protein's interaction partners. For each function of interest and protein, we predict the probability that the protein has such function using Bayesian approaches. Unlike other available approaches for protein annotation in which a protein has or does not have a function of interest, we give a probability for having the function. This probability indicates how confident we are about the prediction. We employ our method to predict protein functions based on "biochemical function," "subcellular location," and "cellular role" for yeast proteins defined in the Yeast Proteome Database (YPD, www.incyte.com), using the protein-protein interaction data from the Munich Information Center for Protein Sequences (MIPS, mips.gsf.de). We show that our approach outperforms other available methods for function prediction based on protein interaction data. The supplementary data is available at www-hto.usc.edu/~msms/ProteinFunction.     

Predicting protein functions with message passing algorithms

MOTIVATION: In the last few years, a growing interest in biology has been shifting toward the problem of optimal information extraction from the huge amount of data generated via large-scale and high-throughput techniques. One of the most relevant issues has recently emerged that of correctly and reliably predicting the functions of a given protein with that of functions exploiting information coming from the whole network of proteins physically interacting with the functionally undetermined one. In the present work, we will refer to an 'observed' protein as the one present in the protein-protein interaction networks published in the literature. METHODS: The method proposed in this paper is based on a message passing algorithm known as Belief Propagation, which accepts the network of protein's physical interactions and a catalog of known protein's functions as input, and returns the probabilities for each unclassified protein of having one chosen function. The implementation of the algorithm allows for fast online analysis, and can easily be generalized into more complex graph topologies taking into account hypergraphs, i.e. complexes of more than two interacting proteins. RESULTS: Benchmarks of our method are the two Saccharomyces cerevisiae protein-protein interaction networks and the Database of Interacting Proteins. The validity of our approach is successfully tested against other available techniques. CONTACT: leone@isiosf.isi.it SUPPLEMENTARY INFORMATION: http://isiosf.isi.it/~pagnani     

Disease gene identification by using graph kernels and Markov random fields

Genes associated with similar diseases are often functionally related. This principle is largely supported by many biological data sources, such as disease phenotype similarities, protein complexes, protein-protein interactions, pathways and gene expression profiles. Integrating multiple types of biological data is an effective method to identify disease genes for many genetic diseases. To capture the gene-disease associations based on biological networks, a kernel-based MRF method is proposed by combining graph kernels and the Markov random field (MRF) method. In the proposed method, three kinds of kernels are employed to describe the overall relationships of vertices in five biological networks, respectively, and a novel weighted MRF method is developed to integrate those data. In addition, an improved Gibbs sampling procedure and a novel parameter estimation method are proposed to generate predictions from the kernel-based MRF method. Numerical experiments are carried out by integrating known gene-disease associations, protein complexes, protein-protein interactions, pathways and gene expression profiles. The proposed kernel-based MRF method is evaluated by the leave-one-out cross validation paradigm, achieving an AUC score of 0.771 when integrating all those biological data in our experiments, which indicates that our proposed method is very promising compared with many existing methods.     

Predicting co-complexed protein pairs from heterogeneous data

Proteins do not carry out their functions alone. Instead, they often act by participating in macromolecular complexes and play different functional roles depending on the other members of the complex. It is therefore interesting to identify co-complex relationships. Although protein complexes can be identified in a high-throughput manner by experimental technologies such as affinity purification coupled with mass spectrometry (APMS), these large-scale datasets often suffer from high false positive and false negative rates. Here, we present a computational method that predicts co-complexed protein pair (CCPP) relationships using kernel methods from heterogeneous data sources. We show that a diffusion kernel based on random walks on the full network topology yields good performance in predicting CCPPs from protein interaction networks. In the setting of direct ranking, a diffusion kernel performs much better than the mutual clustering coefficient. In the setting of SVM classifiers, a diffusion kernel performs much better than a linear kernel. We also show that combination of complementary information improves the performance of our CCPP recognizer. A summation of three diffusion kernels based on two-hybrid, APMS, and genetic interaction networks and three sequence kernels achieves better performance than the sequence kernels or diffusion kernels alone. Inclusion of additional features achieves a still better ROC(50) of 0.937. Assuming a negative-to-positive ratio of 600ratio1, the final classifier achieves 89.3% coverage at an estimated false discovery rate of 10%. Finally, we applied our prediction method to two recently described APMS datasets. We find that our predicted positives are highly enriched with CCPPs that are identified by both datasets, suggesting that our method successfully identifies true CCPPs. An SVM classifier trained from heterogeneous data sources provides accurate predictions of CCPPs in yeast. This computational method thereby provides an inexpensive method for identifying protein complexes that extends and complements high-throughput experimental data.     

An efficient method for protein function annotation based on multilayer protein networks

Background

Accurate annotation of protein functions is still a big challenge for understanding life in the post-genomic era. Many computational methods based on protein-protein interaction (PPI) networks have been proposed to predict the function of proteins. However, the precision of these predictions still needs to be improved, due to the incompletion and noise in PPI networks. Integrating network topology and biological information could improve the accuracy of protein function prediction and may also lead to the discovery of multiple interaction types between proteins. Current algorithms generate a single network, which is archived using a weighted sum of all types of protein interactions.

Method

The influences of different types of interactions on the prediction of protein functions are not the same. To address this, we construct multilayer protein networks (MPN) by integrating PPI networks, the domain of proteins, and information on protein complexes. In the MPN, there is more than one type of connections between pairwise proteins. Different types of connections reflect different roles and importance in protein function prediction. Based on the MPN, we propose a new protein function prediction method, named function prediction based on multilayer protein networks (FP-MPN). Given an un-annotated protein, the FP-MPN method visits each layer of the MPN in turn and generates a set of candidate neighbors with known functions. A set of predicted functions for the testing protein is then formed and all of these functions are scored and sorted. Each layer plays different importance on the prediction of protein functions. A number of top-ranking functions are selected to annotate the unknown protein.

Conclusions

The method proposed in this paper was a better predictor when used on Saccharomyces cerevisiae protein data than other function prediction methods previously used. The proposed FP-MPN method takes different roles of connections in protein function prediction into account to reduce the artificial noise by introducing biological information.

     

Global Network Analysis of Lipid-Raft-Related Proteins Reveals Their Centrality in the Network and Their Roles in Multiple Biological Processes

Lipid rafts are specialized cholesterol-enriched microdomains in the cell membrane. They have been known as a platform for protein-protein interactions and to take part in multiple biological processes. Nevertheless, how lipid rafts influence protein properties at the proteomic level is still an open question for researchers using traditional biochemical approaches. Here, by annotating the lipid raft localization of proteins in human protein-protein interaction networks, we performed a systematic analysis of the function of proteins related to lipid rafts. Our results demonstrated that lipid raft proteins and their interactions were critical for the structure and stability of the whole network, and that the interactions between them were significantly enriched. Furthermore, for each protein in the network, we calculated its “lipid raft dependency (LRD),” which indicates how close it is topologically associated with lipid rafts, and we then uncovered the connection between LRD and protein functions. Proteins with high LRD tended to be essential for mammalian development, and malfunction of these proteins was inclined to cause human diseases. Coordinated with their neighbors, high-LRD proteins participated in multiple biological processes and targeted many pathways in diseases pathogenesis. High-LRD proteins were also found to have tissue specificity of expression. In summary, our network-based analysis denotes that lipid raft proteins have higher centrality in the network, and that lipid-raft-related proteins have multiple functions and are probably concerned with many biological processes in disease development.     

Ranking support vector machine for multiple kernels output combination in protein-protein interaction extraction from biomedical literature

Knowledge about protein-protein interactions (PPIs) unveils the molecular mechanisms of biological processes. However, the volume and content of published biomedical literature on protein interactions is expanding rapidly, making it increasingly difficult for interaction database curators to detect and curate protein interaction information manually. We present a multiple kernel learning-based approach for automatic PPI extraction from biomedical literature. The approach combines the following kernels: feature-based, tree, and graph and combines their output with Ranking support vector machine (SVM). Experimental evaluations show that the features in individual kernels are complementary and the kernel combined with Ranking SVM achieves better performance than those of the individual kernels, equal weight combination and optimal weight combination. Our approach can achieve state-of-the-art performance with respect to the comparable evaluations, with 64.88% F-score and 88.02% AUC on the AImed corpus.     

A Multi-Label Learning Based Kernel Automatic Recommendation Method for Support Vector Machine

Choosing an appropriate kernel is very important and critical when classifying a new problem with Support Vector Machine. So far, more attention has been paid on constructing new kernels and choosing suitable parameter values for a specific kernel function, but less on kernel selection. Furthermore, most of current kernel selection methods focus on seeking a best kernel with the highest classification accuracy via cross-validation, they are time consuming and ignore the differences among the number of support vectors and the CPU time of SVM with different kernels. Considering the tradeoff between classification success ratio and CPU time, there may be multiple kernel functions performing equally well on the same classification problem. Aiming to automatically select those appropriate kernel functions for a given data set, we propose a multi-label learning based kernel recommendation method built on the data characteristics. For each data set, the meta-knowledge data base is first created by extracting the feature vector of data characteristics and identifying the corresponding applicable kernel set. Then the kernel recommendation model is constructed on the generated meta-knowledge data base with the multi-label classification method. Finally, the appropriate kernel functions are recommended to a new data set by the recommendation model according to the characteristics of the new data set. Extensive experiments over 132 UCI benchmark data sets, with five different types of data set characteristics, eleven typical kernels (Linear, Polynomial, Radial Basis Function, Sigmoidal function, Laplace, Multiquadric, Rational Quadratic, Spherical, Spline, Wave and Circular), and five multi-label classification methods demonstrate that, compared with the existing kernel selection methods and the most widely used RBF kernel function, SVM with the kernel function recommended by our proposed method achieved the highest classification performance.     

Faster and more accurate global protein function assignment from protein interaction networks using the MFGO algorithm

MOTIVATION: Predicting protein function accurately is an important issue in the post-genomic era. To achieve this goal, several approaches have been proposed deduce the function of unclassified proteins through sequence similarity, co-expression profiles, and other information. Among these methods, the global optimization method (GOM) is an interesting and powerful tool that assigns functions to unclassified proteins based on their positions in a physical interactions network [Vazquez, A., Flammini, A., Maritan, A. and Vespignani, A. (2003) Global protein function prediction from protein-protein interaction networks, Nat. Biotechnol., 21, 697-700]. To boost both the accuracy and speed of GOM, a new prediction method, MFGO (modified and faster global optimization) is presented in this paper, which employs local optimal repetition method to reduce calculation time, and takes account of topological structure information to achieve a more accurate prediction. CONCLUSION: On four proteins interaction datasets, including Vazquez dataset, YP dataset, DIP-core dataset, and SPK dataset, MFGO was tested and compared with the popular MR (majority rule) and GOM methods. Experimental results confirm MFGO's improvement on both speed and accuracy. Especially, MFGO method has a distinctive advantage in accurately predicting functions for proteins with few neighbors. Moreover, the robustness of the approach was validated both in a dataset containing a high percentage of unknown proteins and a disturbed dataset through random insertion and deletion. The analysis shows that a moderate amount of misplaced interactions do not preclude a reliable function assignment.     

Widely predicting specific protein functions based on protein-protein interaction data and gene expression profile

GESTs (gene expression similarity and taxonomy similarity), a gene functional prediction approach previously proposed by us, is based on gene expression similarity and concept similarity of functional classes defined in Gene Ontology (GO). In this paper, we extend this method to protein-protein interac-tion data by introducing several methods to filter the neighbors in protein interaction networks for a protein of unknown function(s). Unlike other conventional methods, the proposed approach automati-cally selects the most appropriate functional classes as specific as possible during the learning proc-ess, and calls on genes annotated to nearby classes to support the predictions to some small-sized specific classes in GO. Based on the yeast protein-protein interaction information from MIPS and a dataset of gene expression profiles, we assess the performances of our approach for predicting protein functions to “biology process” by three measures particularly designed for functional classes organ-ized in GO. Results show that our method is powerful for widely predicting gene functions with very specific functional terms. Based on the GO database published in December 2004, we predict some proteins whose functions were unknown at that time, and some of the predictions have been confirmed by the new SGD annotation data published in April, 2006.     

Profile-based direct kernels for remote homology detection and fold recognition

MOTIVATION: Protein remote homology detection is a central problem in computational biology. Supervised learning algorithms based on support vector machines are currently one of the most effective methods for remote homology detection. The performance of these methods depends on how the protein sequences are modeled and on the method used to compute the kernel function between them. RESULTS: We introduce two classes of kernel functions that are constructed by combining sequence profiles with new and existing approaches for determining the similarity between pairs of protein sequences. These kernels are constructed directly from these explicit protein similarity measures and employ effective profile-to-profile scoring schemes for measuring the similarity between pairs of proteins. Experiments with remote homology detection and fold recognition problems show that these kernels are capable of producing results that are substantially better than those produced by all of the existing state-of-the-art SVM-based methods. In addition, the experiments show that these kernels, even when used in the absence of profiles, produce results that are better than those produced by existing non-profile-based schemes. AVAILABILITY: The programs for computing the various kernel functions are available on request from the authors.     

Widely predicting specific protein functions based on protein-protein interaction data and gene expression profile

GESTs (gene expression similarity and taxonomy similarity), a gene functional prediction approach previously proposed by us, is based on gene expression similarity and concept similarity of functional classes defined in Gene Ontology (GO). In this paper, we extend this method to protein-protein interaction data by introducing several methods to filter the neighbors in protein interaction networks for a protein of unknown function(s). Unlike other conventional methods, the proposed approach automatically selects the most appropriate functional classes as specific as possible during the learning process, and calls on genes annotated to nearby classes to support the predictions to some small-sized specific classes in GO. Based on the yeast protein-protein interaction information from MIPS and a dataset of gene expression profiles, we assess the performances of our approach for predicting protein functions to “biology process” by three measures particularly designed for functional classes organized in GO. Results show that our method is powerful for widely predicting gene functions with very specific functional terms. Based on the GO database published in December 2004, we predict some proteins whose functions were unknown at that time, and some of the predictions have been confirmed by the new SGD annotation data published in April, 2006.     

Globally predicting protein functions based on co-expressed protein-protein interaction networks and ontology taxonomy similarities

Determining protein functions is an important task in the post-genomic era. Most of the current methods work on some large-sized functional classes selected from functional categorization systems prior to the prediction processes. GESTs, a prediction approach previously proposed by us, is based on gene expression similarity and taxonomy similarity of the functional classes. Unlike many conventional methods, it does not require pre-selecting the functional classes and can predict specific functions for genes according to the functional annotations of their co-expressed genes. In this paper, we extend this method for analyzing protein-protein interaction data. We introduce gene expression data to filter the interacting neighbors of a protein in order to enhance the degree of functional consensus among the neighbors. Using the taxonomy similarity of protein functional classes, the proposed approach can call on the interacting neighbor proteins annotated to nearby classes to support the predictions for an uncharacterized protein, and automatically select the most appropriate small-sized specific functional classes in Gene Ontology (GO) during the learning process. By three measures particularly designed for the functional classes organized in GO, we evaluate the effects of using different taxonomy similarity scores on the prediction performance. Based on the yeast protein-protein interaction data from MIPS and a dataset of gene expression profiles, we show that this method is powerful for predicting protein function to very specific terms. Compared with the other two taxonomy similarity measures used in this study, if we want to achieve higher prediction accuracy with an acceptable specific level (predicted depth), SB-TS measure proposed by us is a reasonable choice for ontology-based functional predictions.     

ClusFCM: an algorithm for predicting protein functions using homologies and protein interactions

We introduce a new algorithm, called ClusFCM, which combines techniques of clustering and fuzzy cognitive maps (FCM) for prediction of protein functions. ClusFCM takes advantage of protein homologies and protein interaction network topology to improve low recall predictions associated with existing prediction methods. ClusFCM exploits the fact that proteins of known function tend to cluster together and deduce functions not only through their direct interaction with other proteins, but also from other proteins in the network. We use ClusFCM to annotate protein functions for Saccharomyces cerevisiae (yeast), Caenorhabditis elegans (worm), and Drosophila melanogaster (fly) using protein-protein interaction data from the General Repository for Interaction Datasets (GRID) database and functional labels from Gene Ontology (GO) terms. The algorithm's performance is compared with four state-of-the-art methods for function prediction--Majority, chi(2) statistics, Markov random field (MRF), and FunctionalFlow--using measures of Matthews correlation coefficient, harmonic mean, and area under the receiver operating characteristic (ROC) curves. The results indicate that ClusFCM predicts protein functions with high recall while not lowering precision. Supplementary information is available at http://www.egr.vcu.edu/cs/dmb/ClusFCM/.     

Improving protein function prediction methods with integrated literature data

Background  

Determining the function of uncharacterized proteins is a major challenge in the post-genomic era due to the problem's complexity and scale. Identifying a protein's function contributes to an understanding of its role in the involved pathways, its suitability as a drug target, and its potential for protein modifications. Several graph-theoretic approaches predict unidentified functions of proteins by using the functional annotations of better-characterized proteins in protein-protein interaction networks. We systematically consider the use of literature co-occurrence data, introduce a new method for quantifying the reliability of co-occurrence and test how performance differs across species. We also quantify changes in performance as the prediction algorithms annotate with increased specificity.     

Prediction of Heterodimeric Protein Complexes from Weighted Protein-Protein Interaction Networks Using Novel Features and Kernel Functions

Since many proteins express their functional activity by interacting with other proteins and forming protein complexes, it is very useful to identify sets of proteins that form complexes. For that purpose, many prediction methods for protein complexes from protein-protein interactions have been developed such as MCL, MCODE, RNSC, PCP, RRW, and NWE. These methods have dealt with only complexes with size of more than three because the methods often are based on some density of subgraphs. However, heterodimeric protein complexes that consist of two distinct proteins occupy a large part according to several comprehensive databases of known complexes. In this paper, we propose several feature space mappings from protein-protein interaction data, in which each interaction is weighted based on reliability. Furthermore, we make use of prior knowledge on protein domains to develop feature space mappings, domain composition kernel and its combination kernel with our proposed features. We perform ten-fold cross-validation computational experiments. These results suggest that our proposed kernel considerably outperforms the naive Bayes-based method, which is the best existing method for predicting heterodimeric protein complexes.     

Bayesian Markov Random Field Analysis for Protein Function Prediction Based on Network Data

Inference of protein functions is one of the most important aims of modern biology. To fully exploit the large volumes of genomic data typically produced in modern-day genomic experiments, automated computational methods for protein function prediction are urgently needed. Established methods use sequence or structure similarity to infer functions but those types of data do not suffice to determine the biological context in which proteins act. Current high-throughput biological experiments produce large amounts of data on the interactions between proteins. Such data can be used to infer interaction networks and to predict the biological process that the protein is involved in. Here, we develop a probabilistic approach for protein function prediction using network data, such as protein-protein interaction measurements. We take a Bayesian approach to an existing Markov Random Field method by performing simultaneous estimation of the model parameters and prediction of protein functions. We use an adaptive Markov Chain Monte Carlo algorithm that leads to more accurate parameter estimates and consequently to improved prediction performance compared to the standard Markov Random Fields method. We tested our method using a high quality S.cereviciae validation network with 1622 proteins against 90 Gene Ontology terms of different levels of abstraction. Compared to three other protein function prediction methods, our approach shows very good prediction performance. Our method can be directly applied to protein-protein interaction or coexpression networks, but also can be extended to use multiple data sources. We apply our method to physical protein interaction data from S. cerevisiae and provide novel predictions, using 340 Gene Ontology terms, for 1170 unannotated proteins and we evaluate the predictions using the available literature.     

Probabilistic protein function prediction from heterogeneous genome-wide data

Dramatic improvements in high throughput sequencing technologies have led to a staggering growth in the number of predicted genes. However, a large fraction of these newly discovered genes do not have a functional assignment. Fortunately, a variety of novel high-throughput genome-wide functional screening technologies provide important clues that shed light on gene function. The integration of heterogeneous data to predict protein function has been shown to improve the accuracy of automated gene annotation systems. In this paper, we propose and evaluate a probabilistic approach for protein function prediction that integrates protein-protein interaction (PPI) data, gene expression data, protein motif information, mutant phenotype data, and protein localization data. First, functional linkage graphs are constructed from PPI data and gene expression data, in which an edge between nodes (proteins) represents evidence for functional similarity. The assumption here is that graph neighbors are more likely to share protein function, compared to proteins that are not neighbors. The functional linkage graph model is then used in concert with protein domain, mutant phenotype and protein localization data to produce a functional prediction. Our method is applied to the functional prediction of Saccharomyces cerevisiae genes, using Gene Ontology (GO) terms as the basis of our annotation. In a cross validation study we show that the integrated model increases recall by 18%, compared to using PPI data alone at the 50% precision. We also show that the integrated predictor is significantly better than each individual predictor. However, the observed improvement vs. PPI depends on both the new source of data and the functional category to be predicted. Surprisingly, in some contexts integration hurts overall prediction accuracy. Lastly, we provide a comprehensive assignment of putative GO terms to 463 proteins that currently have no assigned function.