Progress of 1D protein structure prediction at last

Accuracy of predicting protein secondary structure and solvent accessibility from sequence information has been improved significantly by using information contained in multiple sequence alignments as input to a neural 'network system. For the Asilomar meeting, predictions for 13 proteins were generated automatically using the publicly available prediction method PHD. The results confirm the estimate of 72% three-state prediction accuracy. The fairly accurate predictions of secondary structure segments made the tool useful as a starting point for modeling of higher dimensional aspects of protein structure. © 1995 Wiley-Liss, Inc.     

A simple and fast approach to prediction of protein secondary structure from multiply aligned sequences with accuracy above 70%.

To improve secondary structure predictions in protein sequences, the information residing in multiple sequence alignments of substituted but structurally related proteins is exploited. A database comprised of 70 protein families and a total of 2,500 sequences, some of which were aligned by tertiary structural superpositions, was used to calculate residue exchange weight matrices within alpha-helical, beta-strand, and coil substructures, respectively. Secondary structure predictions were made based on the observed residue substitutions in local regions of the multiple alignments and the largest possible associated exchange weights in each of the three matrix types. Comparison of the observed and predicted secondary structure on a per-residue basis yielded a mean accuracy of 72.2%. Individual alpha-helix, beta-strand, and coil states were respectively predicted at 66.7, and 75.8% correctness, representing a well-balanced three-state prediction. The accuracy level, verified by cross-validation through jack-knife tests on all protein families, dropped, on average, to only 70.9%, indicating the rigor of the prediction procedure. On the basis of robustness, conceptual clarity, accuracy, and executable efficiency, the method has considerable advantage, especially with its sole reliance on amino acid substitutions within structurally related proteins.     

Combining evolutionary information and neural networks to predict protein secondary structure

Using evolutionary information contained in multiple sequence alignments as input to neural networks, secondary structure can be predicted at significantly increased accuracy. Here, we extend our previous three-level system of neural networks by using additional input information derived from multiple alignments. Using a position-specific conservation weight as part of the input increases performance. Using the number of insertions and deletions reduces the tendency for overprediction and increases overall accuracy. Addition of the global amino acid content yields a further improvement, mainly in predicting structural class. The final network system has a sustained overall accuracy of 71.6% in a multiple cross-validation test on 126 unique protein chains. A test on a new set of 124 recently solved protein structures that have no significant sequence similarity to the learning set confirms the high level of accuracy. The average cross-validated accuracy for all 250 sequence-unique chains is above 72%. Using various data sets, the method is compared to alternative prediction methods, some of which also use multiple alignments: the performance advantage of the network system is at least 6 percentage points in three-state accuracy. In addition, the network estimates secondary structure content from multiple sequence alignments about as well as circular dichroism spectroscopy on a single protein and classifies 75% of the 250 proteins correctly into one of four protein structural classes. Of particular practical importance is the definition of a position-specific reliability index. For 40% of all residues the method has a sustained three-state accuracy of 88%, as high as the overall average for homology modelling. A further strength of the method is greatly increased accuracy in predicting the placement of secondary structure segments. © 1994 Wiley-Liss, Inc.     

Easy method to predict solvent accessibility from multiple protein sequence alignments

An easy and uncomplicated method to predict the solvent accessibility state of a site in a multiple protein sequence alignment is described. The approach is based on amino acid exchange and compositional preference matrices for each of three accessibility states: buried, exposed, and intermediate. Calculations utilized a modified version of the 3D―ali databank, a collection of multiple sequence alignments anchored through protein tertiary structural superpositions. The technique achieves the same accuracy as much more complex methods and thus provides such advantages as computational affordability, facile updating, and easily understood residue substitution patterns useful to biochemists involved in protein engineering, design, and structural prediction. The program is available from the authors; and, due to its simplicity, the algorithm can be readily implemented on any system. For a given alignment site, a hand calculation can yield a comparative prediction. Proteins 32:190–199, 1998. © 1998 Wiley-Liss, Inc.     

Factors limiting the performance of prediction-based fold recognition methods

In the past few years, a new generation of fold recognition methods has been developed, in which the classical sequence information is combined with information obtained from secondary structure and, sometimes, accessibility predictions. The results are promising, indicating that this approach may compete with potential-based methods (Rost B et al., 1997, J Mol Biol 270:471-480). Here we present a systematic study of the different factors contributing to the performance of these methods, in particular when applied to the problem of fold recognition of remote homologues. Our results indicate that secondary structure and accessibility prediction methods have reached an accuracy level where they are not the major factor limiting the accuracy of fold recognition. The pattern degeneracy problem is confirmed as the major source of error of these methods. On the basis of these results, we study three different options to overcome these limitations: normalization schemes, mapping of the coil state into the different zones of the Ramachandran plot, and post-threading graphical analysis.     

Predicting protein secondary structure and solvent accessibility with an improved multiple linear regression method

We have improved the multiple linear regression (MLR) algorithm for protein secondary structure prediction by combining it with the evolutionary information provided by multiple sequence alignment of PSI-BLAST. On the CB513 dataset, the three states average overall per-residue accuracy, Q(3), reached 76.4%, while segment overlap accuracy, SOV99, reached 73.2%, using a rigorous jackknife procedure and the strictest reduction of eight states DSSP definition to three states. This represents an improvement of approximately 5% on overall per-residue accuracy compared with previous work. The relative solvent accessibility prediction also benefited from this combination of methods. The system achieved 77.7% average jackknifed accuracy for two states prediction based on a 25% relative solvent accessibility mode, with a Mathews' correlation coefficient of 0.548. The improved MLR secondary structure and relative solvent accessibility prediction server is available at http://spg.biosci.tsinghua.edu.cn/.     

A multiple-template approach to protein threading

Most threading methods predict the structure of a protein using only a single template. Due to the increasing number of solved structures, a protein without solved structure is very likely to have more than one similar template structures. Therefore, a natural question to ask is if we can improve modeling accuracy using multiple templates. This article describes a new multiple-template threading method to answer this question. At the heart of this multiple-template threading method is a novel probabilistic-consistency algorithm that can accurately align a single protein sequence simultaneously to multiple templates. Experimental results indicate that our multiple-template method can improve pairwise sequence-template alignment accuracy and generate models with better quality than single-template models even if they are built from the best single templates (P-value <10(-6)) while many popular multiple sequence/structure alignment tools fail to do so. The underlying reason is that our probabilistic-consistency algorithm can generate accurate multiple sequence/template alignments. In another word, without an accurate multiple sequence/template alignment, the modeling accuracy cannot be improved by simply using multiple templates to increase alignment coverage. Blindly tested on the CASP9 targets with more than one good template structures, our method outperforms all other CASP9 servers except two (Zhang-Server and QUARK of the same group). Our probabilistic-consistency algorithm can possibly be extended to align multiple protein/RNA sequences and structures.     

Real value prediction of solvent accessibility in proteins using multiple sequence alignment and secondary structure

The present study is an attempt to develop a neural network-based method for predicting the real value of solvent accessibility from the sequence using evolutionary information in the form of multiple sequence alignment. In this method, two feed-forward networks with a single hidden layer have been trained with standard back-propagation as a learning algorithm. The Pearson's correlation coefficient increases from 0.53 to 0.63, and mean absolute error decreases from 18.2 to 16% when multiple-sequence alignment obtained from PSI-BLAST is used as input instead of a single sequence. The performance of the method further improves from a correlation coefficient of 0.63 to 0.67 when secondary structure information predicted by PSIPRED is incorporated in the prediction. The final network yields a mean absolute error value of 15.2% between the experimental and predicted values, when tested on two different nonhomologous and nonredundant datasets of varying sizes. The method consists of two steps: (1) in the first step, a sequence-to-structure network is trained with the multiple alignment profiles in the form of PSI-BLAST-generated position-specific scoring matrices, and (2) in the second step, the output obtained from the first network and PSIPRED-predicted secondary structure information is used as an input to the second structure-to-structure network. Based on the present study, a server SARpred (http://www.imtech.res.in/raghava/sarpred/) has been developed that predicts the real value of solvent accessibility of residues for a given protein sequence. We have also evaluated the performance of SARpred on 47 proteins used in CASP6 and achieved a correlation coefficient of 0.68 and a MAE of 15.9% between predicted and observed values.     

Toward the solution of the protein structure prediction problem

Since Anfinsen demonstrated that the information encoded in a protein’s amino acid sequence determines its structure in 1973, solving the protein structure prediction problem has been the Holy Grail of structural biology. The goal of protein structure prediction approaches is to utilize computational modeling to determine the spatial location of every atom in a protein molecule starting from only its amino acid sequence. Depending on whether homologous structures can be found in the Protein Data Bank (PDB), structure prediction methods have been historically categorized as template-based modeling (TBM) or template-free modeling (FM) approaches. Until recently, TBM has been the most reliable approach to predicting protein structures, and in the absence of reliable templates, the modeling accuracy sharply declines. Nevertheless, the results of the most recent community-wide assessment of protein structure prediction experiment (CASP14) have demonstrated that the protein structure prediction problem can be largely solved through the use of end-to-end deep machine learning techniques, where correct folds could be built for nearly all single-domain proteins without using the PDB templates. Critically, the model quality exhibited little correlation with the quality of available template structures, as well as the number of sequence homologs detected for a given target protein. Thus, the implementation of deep-learning techniques has essentially broken through the 50-year-old modeling border between TBM and FM approaches and has made the success of high-resolution structure prediction significantly less dependent on template availability in the PDB library.     

Multiple protein sequence alignment from tertiary structure comparison: assignment of global and residue confidence levels.

An algorithm is presented for the accurate and rapid generation of multiple protein sequence alignments from tertiary structure comparisons. A preliminary multiple sequence alignment is performed using sequence information, which then determines an initial superposition of the structures. A structure comparison algorithm is applied to all pairs of proteins in the superimposed set and a similarity tree calculated. Multiple sequence alignments are then generated by following the tree from the branches to the root. At each branchpoint of the tree, a structure-based sequence alignment and coordinate transformations are output, with the multiple alignment of all structures output at the root. The algorithm encoded in STAMP (STructural Alignment of Multiple Proteins) is shown to give alignments in good agreement with published structural accounts within the dehydrogenase fold domains, globins, and serine proteinases. In order to reduce the need for visual verification, two similarity indices are introduced to determine the quality of each generated structural alignment. Sc quantifies the global structural similarity between pairs or groups of proteins, whereas Pij' provides a normalized measure of the confidence in the alignment of each residue. STAMP alignments have the quality of each alignment characterized by Sc and Pij' values and thus provide a reproducible resource for studies of residue conservation within structural motifs.     

Escher: A new docking procedure applied to the reconstruction of protein tertiary structure

Evaluation of Surface Complementarity, Hydrogen bonding, and Electrostatic interaction in molecular Recognition (ESCHER) is a new docking procedure consisting of three modules that work in series. The first module evaluates the geometric complementarity and produces a set of rough solutions for the docking problem. The second module identifies molecular collisions within those solutions, and the third evaluates their electrostatic complementarity. We describe the algorithm and its application to the docking of cocrystallized protein domains and unbound components of protein-protein complexes. Furthermore, ESCHER has been applied to the reassociation of secondary and supersecondary structure elements. The possibility of applying a docking method to the problem of protein structure prediction is discussed. Proteins 28:556–567, 1997. © 1997 Wiley-Liss, Inc.     

Combining the GOR V algorithm with evolutionary information for protein secondary structure prediction from amino acid sequence

We have modified and improved the GOR algorithm for the protein secondary structure prediction by using the evolutionary information provided by multiple sequence alignments, adding triplet statistics, and optimizing various parameters. We have expanded the database used to include the 513 non-redundant domains collected recently by Cuff and Barton (Proteins 1999;34:508-519; Proteins 2000;40:502-511). We have introduced a variable size window that allowed us to include sequences as short as 20-30 residues. A significant improvement over the previous versions of GOR algorithm was obtained by combining the PSI-BLAST multiple sequence alignments with the GOR method. The new algorithm will form the basis for the future GOR V release on an online prediction server. The average accuracy of the prediction of secondary structure with multiple sequence alignment and full jack-knife procedure was 73.5%. The accuracy of the prediction increases to 74.2% by limiting the prediction to 375 (of 513) sequences having at least 50 PSI-BLAST alignments. The average accuracy of the prediction of the new improved program without using multiple sequence alignments was 67.5%. This is approximately a 3% improvement over the preceding GOR IV algorithm (Garnier J, Gibrat JF, Robson B. Methods Enzymol 1996;266:540-553; Kloczkowski A, Ting K-L, Jernigan RL, Garnier J. Polymer 2002;43:441-449). We have discussed alternatives to the segment overlap (Sov) coefficient proposed by Zemla et al. (Proteins 1999;34:220-223).     

Enhanced protein fold recognition using secondary structure information from NMR.

NMR offers the possibility of accurate secondary structure for proteins that would be too large for structure determination. In the absence of an X-ray crystal structure, this information should be useful as an adjunct to protein fold recognition methods based on low resolution force fields. The value of this information has been tested by adding varying amounts of artificial secondary structure data and threading a sequence through a library of candidate folds. Using a literature test set, the threading method alone has only a one-third chance of producing a correct answer among the top ten guesses. With realistic secondary structure information, one can expect a 60-80% chance of finding a homologous structure. The method has then been applied to examples with published estimates of secondary structure. This implementation is completely independent of sequence homology, and sequences are optimally aligned to candidate structures with gaps and insertions allowed. Unlike work using predicted secondary structure, we test the effect of differing amounts of relatively reliable data.     

Prediction of protein secondary structure content using amino acid composition and evolutionary information

Knowing protein structure and inferring its function from the structure are one of the main issues of computational structural biology, and often the first step is studying protein secondary structure. There have been many attempts to predict protein secondary structure contents. Previous attempts assumed that the content of protein secondary structure can be predicted successfully using the information on the amino acid composition of a protein. Recent methods achieved remarkable prediction accuracy by using the expanded composition information. The overall average error of the most successful method is 3.4%. Here, we demonstrate that even if we only use the simple amino acid composition information alone, it is possible to improve the prediction accuracy significantly if the evolutionary information is included. The idea is motivated by the observation that evolutionarily related proteins share the similar structure. After calculating the homolog-averaged amino acid composition of a protein, which can be easily obtained from the multiple sequence alignment by running PSI-BLAST, those 20 numbers are learned by a multiple linear regression, an artificial neural network and a support vector regression. The overall average error of method by a support vector regression is 3.3%. It is remarkable that we obtain the comparable accuracy without utilizing the expanded composition information such as pair-coupled amino acid composition. This work again demonstrates that the amino acid composition is a fundamental characteristic of a protein. It is anticipated that our novel idea can be applied to many areas of protein bioinformatics where the amino acid composition information is utilized, such as subcellular localization prediction, enzyme subclass prediction, domain boundary prediction, signal sequence prediction, and prediction of unfolded segment in a protein sequence, to name a few.     

A proposed architecture for the central domain of the bacterial enhancer-binding proteins based on secondary structure prediction and fold recognition

The expression of genes transcribed by the RNA polymerase with the alternative sigma factor <r54 (Ecr54) is absolutely dependent on activator proteins that bind to enhancer-like sites, located far upstream from the promoter. These unique prokaryotic proteins, known as enhancer-binding proteins (EBP), mediate open promoter complex formation in a reaction dependent on NTP hydrolysis. The best characterized proteins of this family of regulators are NtrC and Nif A, which activate genes required for ammonia assimilation and nitrogen fixation, respectively. In a recent IRBM course (“Frontiers of protein structure prediction,” IRBM, Pomezia, Italy, 1995; see web site http://www.mrc-cpe.cam.uk/ irbm-course95/), one of us (J.O.) participated in the elaboration of the proposal that the Central domain of the EBPs might adopt the classical mononucleotide-binding fold. This suggestion was based on the results of a new protein fold recognition algorithm (Map) and in the mapping of correlated mutations calculated for the sequence family on the same mononucleotide-binding fold topology. In this work, we present new data that support the previous conclusion. The results from a number of different secondary structure prediction programs suggest that the Central domain could adopt an alfi topology. The fold recognition programs ProFIT 0.9, 3D PROFILE combined with secondary structure prediction, and 123D suggest a mononucleotide-binding fold topology for the Central domain amino acid sequence. Finally, and most importantly, three of five reported residue alterations that impair the Central domain ATPase activity of the Eo-54 activators are mapped to polypeptide regions that might be playing equivalent roles as those involved in nucleotide-binding in the mononucleotide-binding proteins. Furthermore, the known residue substitutions that alter the function of the Ecr54 activators, leaving intact the Central domain ATPase activity, are mapped on a region proposed to play an equivalent role as the effector region of the GTPase superfamily.     

Multiple mapping method: a novel approach to the sequence-to-structure alignment problem in comparative protein structure modeling

A major bottleneck in comparative protein structure modeling is the quality of input alignment between the target sequence and the template structure. A number of alignment methods are available, but none of these techniques produce consistently good solutions for all cases. Alignments produced by alternative methods may be superior in certain segments but inferior in others when compared to each other; therefore, an accurate solution often requires an optimal combination of them. To address this problem, we have developed a new approach, Multiple Mapping Method (MMM). The algorithm first identifies the alternatively aligned regions from a set of input alignments. These alternatively aligned segments are scored using a composite scoring function, which determines their fitness within the structural environment of the template. The best scoring regions from a set of alternative segments are combined with the core part of the alignments to produce the final MMM alignment. The algorithm was tested on a dataset of 1400 protein pairs using 11 combinations of two to four alignment methods. In all cases MMM showed statistically significant improvement by reducing alignment errors in the range of 3 to 17%. MMM also compared favorably over two alignment meta-servers. The algorithm is computationally efficient; therefore, it is a suitable tool for genome scale modeling studies.     

Using genetic algorithms to select most predictive protein features

Many important characteristics of proteins such as biochemical activity and subcellular localization present a challenge to machine-learning methods: it is often difficult to encode the appropriate input features at the residue level for the purpose of making a prediction for the entire protein. The problem is usually that the biophysics of the connection between a machine-learning method's input (sequence feature) and its output (observed phenomenon to be predicted) remains unknown; in other words, we may only know that a certain protein is an enzyme (output) without knowing which region may contain the active site residues (input). The goal then becomes to dissect a protein into a vast set of sequence-derived features and to correlate those features with the desired output. We introduce a framework that begins with a set of global sequence features and then vastly expands the feature space by generically encoding the coexistence of residue-based features. It is this combination of individual features, that is the step from the fractions of serine and buried (input space 20 + 2) to the fraction of buried serine (input space 20 * 2) that implicitly shifts the search space from global feature inputs to features that can capture very local evidence such as a the individual residues of a catalytic triad. The vast feature space created is explored by a genetic algorithm (GA) paired with neural networks and support vector machines. We find that the GA is critical for selecting combinations of features that are neither too general resulting in poor performance, nor too specific, leading to overtraining. The final framework manages to effectively sample a feature space that is far too large for exhaustive enumeration. We demonstrate the power of the concept by applying it to prediction of protein enzymatic activity.     

A contact energy function considering residue hydrophobic environment and its application in protein fold recognition

The three-dimensional （3D） structure prediction of proteins ：is an important task in bioinformatics. Finding energy functions that can better represent residue-residue and residue-solvent interactions is a crucial way to improve the prediction accu- racy. The widely used contact energy functions mostly only consider the contact frequency between different types of residues; however, we find that the contact frequency also relates to the residue hydrophobic environment. Accordingly, we present an improved contact energy function to integrate the two factors, which can reflect the influence of hydrophobic interaction on the stabilization of protein 3D structure more effectively. Furthermore, a fold recognition （threading） approach based on this energy function is developed. The testing results obtained with 20 randomly selected proteins demonstrate that, compared with common contact energy functions, the proposed energy function can improve the accuracy of the fold template prediction from 20% to 50%, and can also improve the accuracy of the sequence-template alignment from 35% to 65%.     

Analysis of distance-based protein structure prediction by deep learning in CASP13

This paper reports the CASP13 results of distance-based contact prediction, threading, and folding methods implemented in three RaptorX servers, which are built upon the powerful deep convolutional residual neural network (ResNet) method initiated by us for contact prediction in CASP12. On the 32 CASP13 FM (free-modeling) targets with a median multiple sequence alignment (MSA) depth of 36, RaptorX yielded the best contact prediction among 46 groups and almost the best 3D structure modeling among all server groups without time-consuming conformation sampling. In particular, RaptorX achieved top L/5, L/2, and L long-range contact precision of 70%, 58%, and 45%, respectively, and predicted correct folds (TMscore > 0.5) for 18 of 32 targets. Further, RaptorX predicted correct folds for all FM targets with >300 residues (T0950-D1, T0969-D1, and T1000-D2) and generated the best 3D models for T0950-D1 and T0969-D1 among all groups. This CASP13 test confirms our previous findings: (a) predicted distance is more useful than contacts for both template-based and free modeling; and (b) structure modeling may be improved by integrating template and coevolutionary information via deep learning. This paper will discuss progress we have made since CASP12, the strength and weakness of our methods, and why deep learning performed much better in CASP13.     

GLASS: A tool to visualize protein structure prediction data in three dimensions and evaluate their consistency

When a protein sequence does not share any significant sequence similarity with a protein of known structure, homology modeling cannot be applied. However, many novel and interesting methods, such as secondary structure prediction, fold recognition, and prediction of long-range interactions, are being developed and have been shown to be reasonably successful in predicting protein structures from sequence data and evolutionary information. The a priori evaluation of the correctness of a prediction obtained by one of these methods is however often problematic. Consequently, it is important to use all available information provided by as many different methods as possible and all the available experimental data about the protein of interest, since the consistency of the results is indicative of the reliability of the prediction. Hence the need has arisen for suitable tools able to compare results provided by different methods and evaluate their consistency. We have therefore constructed GLASS, a general platform to read, visualize, compare, and evaluate prediction results from many different sources and to project these prediction results into three dimensions. In addition, GLASS allows the comparison of selected parameters calculated for a model with the distribution observed in real protein structures, thus providing an easy way to test new methods for evaluating the likelihood of different structural models. GLASS can be considered as a “workbench” for structural predictions useful to both experimentalists and theoreticians. Proteins 30:339–351, 1998. © 1998 Wiley-Liss, Inc.