Predicting protein secondary structure with probabilistic schemata of evolutionarily derived information

Abstract

We demonstrate the applicability of our previously developed Bayesian probabilistic approach for predicting residue solvent accessibility to the problem of predicting secondary structure. Using only single-sequence data, this method achieves a three-state accuracy of 67% over a database of 473 non-homologous proteins. This approach is more amenable to inspection and less likely to overlearn specifics of a dataset than "black box" methods such as neural networks. It is also conceptually simpler and less computationally costly.We also introduce a novel method for representing and incorporating multiple-sequence alignment information within the prediction algorithm, achieving 72% accuracy over a dataset of 304 non-homologous proteins. This is accomplished by creating a statistical model of the evolutionarily derived correlations between patterns of amino acid substitution and local protein structure. This model consists of parameter vectors, termed "substitution schemata," which probabilistically encode the structure-based heterogeneity in the distributions of amino acid substitutions found in alignments of homologous proteins. The model is optimized for structure prediction by maximizing the mutual information between the set of schemata and the database of secondary structures.Unlike "expert heuristic" methods, this approach has been demonstrated to work well over large datasets. Unlike the opaque neural network algorithms, this approach is physicochemically intelligible. Moreover, the model optimization procedure, the formalism for predicting one-dimensional structural features, and our previously developed method for tertiary structure recognition all share a common Bayesian probabilistic basis. This consistency starkly contrasts with the hybrid and ad hoc nature of methods that have dominated this field in recent years.Keywords: Bayesian statistics; evolutionary information; mutual information; probabilistic schemata; secondary structure prediction The prediction of protein secondary structure by a number of methods has benefitted from the use of aligned sets of homologous proteins. The patterns of conservation and variation in amino acid residue substitutions at a particular site in a protein convey implicit information about the long-range interactions involved in determining the local structure at that site. Various techniques have been devised to use this information to raise the three-state accuracies to around 70-72%. While this summary statistic is universally reported, a broader evaluation of prediction performance would consider both the practical and scientific value of the prediction scheme in terms of its statistical performance, physicochemical interpretability, and general applicability (reproducibility and robustness). Until now, the success of multiple-sequence alignment-based secondary structure prediction methods has been restricted to one or two of these attributes. The most common technique for using multiple-sequence alignments is the consensus method. This "signal averaging" appro...

Protein secondary structure: entropy, correlations and prediction

“…Instead of a large primary structure window, we use short, overlapping secondary structure windows. Similar models, with similar assumptions, can be found in the work of Thompson and Goldstein (1997) and Schmidler et al (2000). The probability, P (S|R), of a secondary structure sequence S, given the primary sequence, R, can be rewritten using Bayes' rule as…”

Bayesian probabilistic approach for predicting backbone structures in terms of protein blocks

“…9,38,39 Relative to neural network approach, the scores directly reflect the sequence information content expressed from the amino acid composition in every site of the sequence window. It is not only limited to a PBs ordering.…”

“…A thousand different prediction algorithms have been developed, e.g., statistical methods like the pioneer GOR 3,4 or neural networks like the well-known PHD 5 and the more recent work of Chandonia and Karplus. 6,7 The accuracy of these works were strongly increased with the addition of the multiple sequences alignment in the neural networks, 8 probabilistic approach, 9 or computational informative encoding. 10 The increase in the entries in the biologic databases may permit an increase in the prediction rate.…”

“…So, by using Bayes theorem, these probabilities may be further used to predict the structural motifs able to be adopted by a given protein chain. The Bayesian probabilistic approach is largely applied in this type of study, for instance predicting solvent accessibility, 38 secondary structure, 9 or characterization of biologic pathways from sequences to functions. 39 In this study, we have worked in different aspects to improve the PB prediction: 1 Protein Block-n Sequences: Associating one protein block with one class of sequences is a restrictive point of view.…”

Selective Inhibition of the Collagenase Activity of Cathepsin K

“…Interestingly, a size of at least 15 Glu residues (E15) is required to inhibit complex formation. A prediction of the secondary structure for poly-Glu of varying length based on an implementation of the Bayesian prediction formalism in the modeling program MOE (20) suggests a helix as a secondary structure for molecules sized E15 and above, which agrees with inhibitory potencies for complex inhibition. It can be speculated that the formation of a helix optimally organizes the spatial orientation of carboxylate groups for an efficient interaction with the Lys and Arg residues of cathepsin K, thus resulting in a potent complex inhibition.…”