Predicting enzyme class from protein structure without alignments

Methods for predicting protein function from structure are becoming more important as the rate at which structures are solved increases more rapidly than experimental knowledge. As a result, protein structures now frequently lack functional annotations. The majority of methods for predicting protein function are reliant upon identifying a similar protein and transferring its annotations to the query protein. This method fails when a similar protein cannot be identified, or when any similar proteins identified also lack reliable annotations. Here, we describe a method that can assign function from structure without the use of algorithms reliant upon alignments. Using simple attributes that can be calculated from any crystal structure, such as secondary structure content, amino acid propensities, surface properties and ligands, we describe each enzyme in a non-redundant set. The set is split according to Enzyme Classification (EC) number. We combine the predictions of one-class versus one-class support vector machine models to make overall assignments of EC number to an accuracy of 35% with the top-ranked prediction, rising to 60% accuracy with the top two ranks. In doing so we demonstrate the utility of simple structural attributes in protein function prediction and shed light on the link between structure and function. We apply our methods to predict the function of every currently unclassified protein in the Protein Data Bank.     

Prediction of enzymes and non-enzymes from protein sequences based on sequence derived features and PSSM matrix using artificial neural network

The problem of predicting the enzymes and non-enzymes from the protein sequence information is still an open problem in bioinformatics. It is further becoming more important as the number of sequenced information grows exponentially over time. We describe a novel approach for predicting the enzymes and non-enzymes from its amino-acid sequence using artificial neural network (ANN). Using 61 sequence derived features alone we have been able to achieve 79 percent correct prediction of enzymes/non-enzymes (in the set of 660 proteins). For the complete set of 61 parameters using 5-fold cross-validated classification, ANN model reveal a superior model (accuracy = 78.79 plus or minus 6.86 percent, Q(pred) = 74.734 plus or minus 17.08 percent, sensitivity = 84.48 plus or minus 6.73 percent, specificity = 77.13 plus or minus 13.39 percent). The second module of ANN is based on PSSM matrix. Using the same 5-fold cross-validation set, this ANN model predicts enzymes/non-enzymes with more accuracy (accuracy = 80.37 plus or minus 6.59 percent, Q(pred) = 67.466 plus or minus 12.41 percent, sensitivity = 0.9070 plus or minus 3.37 percent, specificity = 74.66 plus or minus 7.17 percent).     

A top-down approach to classify enzyme functional classes and sub-classes using random forest

Advancements in sequencing technologies have witnessed an exponential rise in the number of newly found enzymes. Enzymes are proteins that catalyze bio-chemical reactions and play an important role in metabolic pathways. Commonly, function of such enzymes is determined by experiments that can be time consuming and costly. Hence, a need for a computing method is felt that can distinguish protein enzyme sequences from those of non-enzymes and reliably predict the function of the former. To address this problem, approaches that cluster enzymes based on their sequence and structural similarity have been presented. But, these approaches are known to fail for proteins that perform the same function and are dissimilar in their sequence and structure. In this article, we present a supervised machine learning model to predict the function class and sub-class of enzymes based on a set of 73 sequence-derived features. The functional classes are as defined by International Union of Biochemistry and Molecular Biology. Using an efficient data mining algorithm called random forest, we construct a top-down three layer model where the top layer classifies a query protein sequence as an enzyme or non-enzyme, the second layer predicts the main function class and bottom layer further predicts the sub-function class. The model reported overall classification accuracy of 94.87% for the first level, 87.7% for the second, and 84.25% for the bottom level. Our results compare very well with existing methods, and in many cases report better performance. Using feature selection methods, we have shown the biological relevance of a few of the top rank attributes.     

On the accuracy of homology modeling and sequence alignment methods applied to membrane proteins

In this study, we investigate the extent to which techniques for homology modeling that were developed for water-soluble proteins are appropriate for membrane proteins as well. To this end we present an assessment of current strategies for homology modeling of membrane proteins and introduce a benchmark data set of homologous membrane protein structures, called HOMEP. First, we use HOMEP to reveal the relationship between sequence identity and structural similarity in membrane proteins. This analysis indicates that homology modeling is at least as applicable to membrane proteins as it is to water-soluble proteins and that acceptable models (with C alpha-RMSD values to the native of 2 A or less in the transmembrane regions) may be obtained for template sequence identities of 30% or higher if an accurate alignment of the sequences is used. Second, we show that secondary-structure prediction algorithms that were developed for water-soluble proteins perform approximately as well for membrane proteins. Third, we provide a comparison of a set of commonly used sequence alignment algorithms as applied to membrane proteins. We find that high-accuracy alignments of membrane protein sequences can be obtained using state-of-the-art profile-to-profile methods that were developed for water-soluble proteins. Improvements are observed when weights derived from the secondary structure of the query and the template are used in the scoring of the alignment, a result which relies on the accuracy of the secondary-structure prediction of the query sequence. The most accurate alignments were obtained using template profiles constructed with the aid of structural alignments. In contrast, a simple sequence-to-sequence alignment algorithm, using a membrane protein-specific substitution matrix, shows no improvement in alignment accuracy. We suggest that profile-to-profile alignment methods should be adopted to maximize the accuracy of homology models of membrane proteins.     

Protein homology detection and fold inference through multiple alignment entropy profiles

Homology detection and protein structure prediction are central themes in bioinformatics. Establishment of relationship between protein sequences or prediction of their structure by sequence comparison methods finds limitations when there is low sequence similarity. Recent works demonstrate that the use of profiles improves homology detection and protein structure prediction. Profiles can be inferred from protein multiple alignments using different approaches. The "Conservatism-of-Conservatism" is an effective profile analysis method to identify structural features between proteins having the same fold but no detectable sequence similarity. The information obtained from protein multiple alignments varies according to the amino acid classification employed to calculate the profile. In this work, we calculated entropy profiles from PSI-BLAST-derived multiple alignments and used different amino acid classifications summarizing almost 500 different attributes. These entropy profiles were converted into pseudocodes which were compared using the FASTA program with an ad-hoc matrix. We tested the performance of our method to identify relationships between proteins with similar fold using a nonredundant subset of sequences having less than 40% of identity. We then compared our results using Coverage Versus Error per query curves, to those obtained by methods like PSI-BLAST, COMPASS and HHSEARCH. Our method, named HIP (Homology Identification with Profiles) presented higher accuracy detecting relationships between proteins with the same fold. The use of different amino acid classifications reflecting a large number of amino acid attributes, improved the recognition of distantly related folds. We propose the use of pseudocodes representing profile information as a fast and powerful tool for homology detection, fold assignment and analysis of evolutionary information enclosed in protein profiles.     

Structure-based function inference using protein family-specific fingerprints

We describe a method to assign a protein structure to a functional family using family-specific fingerprints. Fingerprints represent amino acid packing patterns that occur in most members of a family but are rare in the background, a nonredundant subset of PDB; their information is additional to sequence alignments, sequence patterns, structural superposition, and active-site templates. Fingerprints were derived for 120 families in SCOP using Frequent Subgraph Mining. For a new structure, all occurrences of these family-specific fingerprints may be found by a fast algorithm for subgraph isomorphism; the structure can then be assigned to a family with a confidence value derived from the number of fingerprints found and their distribution in background proteins. In validation experiments, we infer the function of new members added to SCOP families and we discriminate between structurally similar, but functionally divergent TIM barrel families. We then apply our method to predict function for several structural genomics proteins, including orphan structures. Some predictions have been corroborated by other computational methods and some validated by subsequent functional characterization.     

Analysis and prediction of functional sub-types from protein sequence alignments

The increasing number and diversity of protein sequence families requires new methods to define and predict details regarding function. Here, we present a method for analysis and prediction of functional sub-types from multiple protein sequence alignments. Given an alignment and set of proteins grouped into sub-types according to some definition of function, such as enzymatic specificity, the method identifies positions that are indicative of functional differences by comparison of sub-type specific sequence profiles, and analysis of positional entropy in the alignment. Alignment positions with significantly high positional relative entropy correlate with those known to be involved in defining sub-types for nucleotidyl cyclases, protein kinases, lactate/malate dehydrogenases and trypsin-like serine proteases. We highlight new positions for these proteins that suggest additional experiments to elucidate the basis of specificity. The method is also able to predict sub-type for unclassified sequences. We assess several variations on a prediction method, and compare them to simple sequence comparisons. For assessment, we remove close homologues to the sequence for which a prediction is to be made (by a sequence identity above a threshold). This simulates situations where a protein is known to belong to a protein family, but is not a close relative of another protein of known sub-type. Considering the four families above, and a sequence identity threshold of 30 %, our best method gives an accuracy of 96 % compared to 80 % obtained for sequence similarity and 74 % for BLAST. We describe the derivation of a set of sub-type groupings derived from an automated parsing of alignments from PFAM and the SWISSPROT database, and use this to perform a large-scale assessment. The best method gives an average accuracy of 94 % compared to 68 % for sequence similarity and 79 % for BLAST. We discuss implications for experimental design, genome annotation and the prediction of protein function and protein intra-residue distances.     

Using support vector machines to distinguish enzymes: Approached by incorporating wavelet transform

The enzymatic attributes of newly found protein sequences are usually determined either by biochemical analysis of eukaryotic and prokaryotic genomes or by microarray chips. These experimental methods are both time-consuming and costly. With the explosion of protein sequences registered in the databanks, it is highly desirable to develop an automated method to identify whether a given new sequence belongs to enzyme or non-enzyme. The discrete wavelet transform (DWT) and support vector machine (SVM) have been used in this study for distinguishing enzyme structures from non-enzymes. The networks have been trained and tested on two datasets of proteins with different wavelet basis functions, decomposition scales and hydrophobicity data types. Maximum accuracy has been obtained using SVM with a wavelet function of Bior2.4, a decomposition scale j=5, and Kyte-Doolittle hydrophobicity scales. The results obtained by the self-consistency test, jackknife test and independent dataset test are encouraging, which indicates that the proposed method can be employed as a useful assistant technique for distinguishing enzymes from non-enzymes.     

Structural similarity to link sequence space: new potential superfamilies and implications for structural genomics

The current pace of structural biology now means that protein three-dimensional structure can be known before protein function, making methods for assigning homology via structure comparison of growing importance. Previous research has suggested that sequence similarity after structure-based alignment is one of the best discriminators of homology and often functional similarity. Here, we exploit this observation, together with a merger of protein structure and sequence databases, to predict distant homologous relationships. We use the Structural Classification of Proteins (SCOP) database to link sequence alignments from the SMART and Pfam databases. We thus provide new alignments that could not be constructed easily in the absence of known three-dimensional structures. We then extend the method of Murzin (1993b) to assign statistical significance to sequence identities found after structural alignment and thus suggest the best link between diverse sequence families. We find that several distantly related protein sequence families can be linked with confidence, showing the approach to be a means for inferring homologous relationships and thus possible functions when proteins are of known structure but of unknown function. The analysis also finds several new potential superfamilies, where inspection of the associated alignments and superimpositions reveals conservation of unusual structural features or co-location of conserved amino acids and bound substrates. We discuss implications for Structural Genomics initiatives and for improvements to sequence comparison methods.     

Prediction of human protein function from post-translational modifications and localization features

We have developed an entirely sequence-based method that identifies and integrates relevant features that can be used to assign proteins of unknown function to functional classes, and enzyme categories for enzymes. We show that strategies for the elucidation of protein function may benefit from a number of functional attributes that are more directly related to the linear sequence of amino acids, and hence easier to predict, than protein structure. These attributes include features associated with post-translational modifications and protein sorting, but also much simpler aspects such as the length, isoelectric point and composition of the polypeptide chain.     

Structure-based evaluation of sequence comparison and fold recognition alignment accuracy

The biological role, biochemical function, and structure of uncharacterized protein sequences is often inferred from their similarity to known proteins. A constant goal is to increase the reliability, sensitivity, and accuracy of alignment techniques to enable the detection of increasingly distant relationships. Development, tuning, and testing of these methods benefit from appropriate benchmarks for the assessment of alignment accuracy.Here, we describe a benchmark protocol to estimate sequence-to-sequence and sequence-to-structure alignment accuracy. The protocol consists of structurally related pairs of proteins and procedures to evaluate alignment accuracy over the whole set. The set of protein pairs covers all the currently known fold types. The benchmark is challenging in the sense that it consists of proteins lacking clear sequence similarity.Correct target alignments are derived from the three-dimensional structures of these pairs by rigid body superposition. An evaluation engine computes the accuracy of alignments obtained from a particular algorithm in terms of alignment shifts with respect to the structure derived alignments. Using this benchmark we estimate that the best results can be obtained from a combination of amino acid residue substitution matrices and knowledge-based potentials.     

Active site identification through geometry-based and sequence profile-based calculations: burial of catalytic clefts

Electrostatics calculations with proteins that are uniformly charged over volume can aid enzyme/non-enzyme discrimination. For known enzymes, such methods locate active sites to within 5% on the enzyme surface, in 77% of a test set. We now report that removing the dielectric boundary improves active site location to 80%, with optimal discrimination between enzymes and non-enzymes of around 80% specificity and 80% sensitivity. This calculation quantifies burial of solvent-accessible regions. Many of the true enzymes incorrectly assigned as non-enzymes have active sites at subunit boundaries. These are missed in monomer-based calculations. Catalytic and non-catalytic antibodies are studied in this context of active/binding site burial. Whilst catalytic antibodies, on average, have marginally higher active site burial than non-catalytic antibodies, these values are generally smaller than for non-antibody enzymes, possibly contributing to their relatively low turnover. Prediction of active site location improves further when sequence profile-based weights replace the uniform charge distribution, so that a combination of burial and amino acid conservation is assessed. Accuracy rises to 93% of active sites to within 5%, in the test set, for the optimal profile weights scheme. The equivalent value in a separate validation set is 89% to within 5%. Enzyme/non-enzyme and enzyme functional site predictions are made for structural genomics proteins, suggesting that a substantial majority of these are non-enzymes.     

FrankSum: new feature selection method for protein function prediction

In the study of in silico functional genomics, improving the performance of protein function prediction is the ultimate goal for identifying proteins associated with defined cellular functions. The classical prediction approach is to employ pairwise sequence alignments. However this method often faces difficulties when no statistically significant homologous sequences are identified. An alternative way is to predict protein function from sequence-derived features using machine learning. In this case the choice of possible features which can be derived from the sequence is of vital importance to ensure adequate discrimination to predict function. In this paper we have successfully selected biologically significant features for protein function prediction. This was performed using a new feature selection method (FrankSum) that avoids data distribution assumptions, uses a data independent measurement (p-value) within the feature, identifies redundancy between features and uses an appropriate ranking criterion for feature selection. We have shown that classifiers generated from features selected by FrankSum outperforms classifiers generated from full feature sets, randomly selected features and features selected from the Wrapper method. We have also shown the features are concordant across all species and top ranking features are biologically informative. We conclude that feature selection is vital for successful protein function prediction and FrankSum is one of the feature selection methods that can be applied successfully to such a domain.     

A word-oriented approach to alignment validation

MOTIVATION: Multiple sequence alignment at the level of whole proteomes requires a high degree of automation, precluding the use of traditional validation methods such as manual curation. Since evolutionary models are too general to describe the history of each residue in a protein family, there is no single algorithm/model combination that can yield a biologically or evolutionarily optimal alignment. We propose a 'shotgun' strategy where many different algorithms are used to align the same family, and the best of these alignments is then chosen with a reliable objective function. We present WOOF, a novel 'word-oriented' objective function that relies on the identification and scoring of conserved amino acid patterns (words) between pairs of sequences. RESULTS: Tests on a subset of reference protein alignments from BAliBASE showed that WOOF tended to rank the (manually curated) reference alignment highest among 1060 alternative (automatically generated) alignments for a majority of protein families. Among the automated alignments, there was a strong positive relationship between the WOOF score and similarity to the reference alignment. The speed of WOOF and its independence from explicit considerations of three-dimensional structure make it an excellent tool for analyzing large numbers of protein families. AVAILABILITY: On request from the authors.     

SnapDRAGON: a method to delineate protein structural domains from sequence data

We describe a method to identify protein domain boundaries from sequence information alone based on the assumption that hydrophobic residues cluster together in space. SnapDRAGON is a suite of programs developed to predict domain boundaries based on the consistency observed in a set of alternative ab initio three-dimensional (3D) models generated for a given protein multiple sequence alignment. This is achieved by running a distance geometry-based folding technique in conjunction with a 3D-domain assignment algorithm. The overall accuracy of our method in predicting the number of domains for a non-redundant data set of 414 multiple alignments, representing 185 single and 231 multiple-domain proteins, is 72.4 %. Using domain linker regions observed in the tertiary structures associated with each query alignment as the standard of truth, inter-domain boundary positions are delineated with an accuracy of 63.9 % for proteins comprising continuous domains only, and 35.4 % for proteins with discontinuous domains. Overall, domain boundaries are delineated with an accuracy of 51.8 %. The prediction accuracy values are independent of the pair-wise sequence similarities within each of the alignments. These results demonstrate the capability of our method to delineate domains in protein sequences associated with a wide variety of structural domain organisation.     

Detecting thermophilic proteins through selecting amino acid and dipeptide composition features

Detecting thermophilic proteins is an important task for designing stable protein engineering in interested temperatures. In this work, we develop a simple but efficient method to classify thermophilic proteins from mesophilic ones using the amino acid and dipeptide compositions. Since most of the amino acid and dipeptide compositions are redundant, we propose a new forward floating selection technique to select only a useful subset of these compositions as features for support vector machine-based classification. We test the proposed method on a benchmark data set of 915 thermophilic and 793 mesophilic proteins. The results show that our method using 28 amino acid and dipeptide compositions achieves an accuracy rate of 93.3% evaluated by the jackknife cross-validation test, which is higher not only than the existing methods but also than using all amino acid and dipeptide compositions.     

Homology modeling using parametric alignment ensemble generation with consensus and energy-based model selection

The accuracy of a homology model based on the structure of a distant relative or other topologically equivalent protein is primarily limited by the quality of the alignment. Here we describe a systematic approach for sequence-to-structure alignment, called ‘K*Sync’, in which alignments are generated by dynamic programming using a scoring function that combines information on many protein features, including a novel measure of how obligate a sequence region is to the protein fold. By systematically varying the weights on the different features that contribute to the alignment score, we generate very large ensembles of diverse alignments, each optimal under a particular constellation of weights. We investigate a variety of approaches to select the best models from the ensemble, including consensus of the alignments, a hydrophobic burial measure, low- and high-resolution energy functions, and combinations of these evaluation methods. The effect on model quality and selection resulting from loop modeling and backbone optimization is also studied. The performance of the method on a benchmark set is reported and shows the approach to be effective at both generating and selecting accurate alignments. The method serves as the foundation of the homology modeling module in the Robetta server.     

Enzyme family classification by support vector machines

One approach for facilitating protein function prediction is to classify proteins into functional families. Recent studies on the classification of G-protein coupled receptors and other proteins suggest that a statistical learning method, Support vector machines (SVM), may be potentially useful for protein classification into functional families. In this work, SVM is applied and tested on the classification of enzymes into functional families defined by the Enzyme Nomenclature Committee of IUBMB. SVM classification system for each family is trained from representative enzymes of that family and seed proteins of Pfam curated protein families. The classification accuracy for enzymes from 46 families and for non-enzymes is in the range of 50.0% to 95.7% and 79.0% to 100% respectively. The corresponding Matthews correlation coefficient is in the range of 54.1% to 96.1%. Moreover, 80.3% of the 8,291 correctly classified enzymes are uniquely classified into a specific enzyme family by using a scoring function, indicating that SVM may have certain level of unique prediction capability. Testing results also suggest that SVM in some cases is capable of classification of distantly related enzymes and homologous enzymes of different functions. Effort is being made to use a more comprehensive set of enzymes as training sets and to incorporate multi-class SVM classification systems to further enhance the unique prediction accuracy. Our results suggest the potential of SVM for enzyme family classification and for facilitating protein function prediction. Our software is accessible at http://jing.cz3.nus.edu.sg/cgi-bin/svmprot.cgi.     

Predicting enzyme family classes by hybridizing gene product composition and pseudo-amino acid composition

A new method has been developed to predict the enzymatic attribute of proteins by hybridizing the gene product composition and pseudo amino acid composition. As a demonstration, a working dataset was generated with a cutoff of 60% sequence identity to avoid redundancy and bias in statistical prediction. The dataset thus constructed contains 39989 protein sequences, of which 27469 are non-enzymes and 12520 enzymes that were further classified into 6 enzyme family classes according to their 6 main EC (Enzyme Commission) numbers (2314 are oxidoreductases, 3653 transferases, 3246 hydrolases, 1307 lyases, 676 isomerases, and 1324 ligases). The overall success rate by the jackknife test for the identification between enzyme and non-enzyme was 94%, and that for the identification among the 6 enzyme family classes was 98%. It is anticipated that, with the rapid increase of protein sequences entering into databanks, the current method will become a useful automated tool in identifying the enzymatic attribute of a newly found protein sequence.