Protein secondary structure prediction

Since these studies used local encoding schemes which utilized limited correlation information between residues, little or no improvement was shown by using a multilayered network with hidden units (Qian Sejnowski, 1988 Stolorz et al., 1992 Fariselli et air, 1993). A performance ceiling of about 65% three-state accuracy was observed in these networks. The results were only marginally more accurate than a simplistic Bayesian statistical method that assumed independent probabilities of amino acid residues (Stolorz et al., 1992). 

It has been observed that the use of protein tertiary structural class improved the accuracy for a 2-state secondary structure prediction (Kneller et al., 1990). A modular network architecture was proposed using separate networks (i.e., a- or P-type network) for classification of different secondary structures (Sasagawa Tajima, 1993). Recently, Chandonia Karplus (1995) trained a pair of neural networks to predict the protein secondary structure and the structural class respectively. Using predicted class information, the secondary structure prediction network realized a small increase in accuracy. 

Fariselli et al. (1993) extended the neural network design for secondary structure prediction of globular proteins to the prediction of membrane proteins. The result was better than those obtained with statistical methods used for membrane proteins. Their findings also indicated that regular patterns of secondary structures are common for globular and membrane proteins. 

Vivarelli et al. (1995) used a hybrid system that combined a local genetic algorithm (LGA) and neural networks for the protein secondary structure prediction. The LGA, a version of the genetic algorithms (GAs), was particularly suitable for parallel computational architectures. Although the LGA was effective in selecting different 

Backpropagation Neural Networks NN1 (Sequence-Structure) --------- NN2 (Structure-Structure) 

---

JE Gibrat, J Garnier, B Robson. Eurther developments of protein secondary structure prediction using information theory. New parameters and consideration of residue pairs. J Mol Biol 198 425-443, 1987. 

LH Holley, M Karplus. Protein secondary structure prediction with a neural network. Proc Natl Acad Sci USA 86 152-156, 1989. 

JM Chandoma, M Karplus. The importance of larger data sets for protein secondary structure prediction with neural networks. Protein Sci 5 768-774, 1996. 

Barton, G.J. Protein secondary structure prediction. Curr. Opin. Struct. Biol. 5 372-376, 1995. 

Rost, B., Sander, C., Schneider, R. Redefining the goals of protein secondary structure prediction. /. Mol. Biol. 

Jones, D. T. (1999) Protein secondary structure prediction based on position-specific scoring matrices. J. Mol. Biol. 292, 195-202. 

A number of servers offer various methods to predict secondary structures of proteins. Secondary structure prediction of ExPASy Proteomic tools (http //... 

TABLE 12.1. Web Servers for Protein Secondary Structure Prediction... 

Kneller DG, Cohen FE, Langridge R. Improvements in protein secondary structure prediction hy an enhanced neural network. J Mol Biol 1990 214 171-182. 

J.-F. GibraL, J. Gamier, and B. Robson. Further developments of protein secondary structure prediction usina information theonr. J. MoL BioL 796 425-443 (1987). 

Source Data from Liljas A, Rossmann MG. X-ray studies of protein interactions. Annu Rev Biochem 43 475-505, 1974 Argos P, Schwarz JS, Schwarz J. An assessment of protein secondary structure prediction methods based on amino acid sequence. Biochim Biophys Acta 439 261-273, 1976. 

Another frequently used global information that covers protein context is the residue frequencies. The composition is often calibrated with that from the database as in Gamier et al. (1996), where only observed frequencies of amino acids and amino acid pairs are used for protein secondary structure prediction. 

In protein secondary structure prediction, where a three-category (a, P, and coil or loop) prediction is made, the accuracy can be measured by a 3 x 3 accuracy table, as in Rost and Sander (1993). 

Table 8.1 A 3x3 accuracy matrix for evaluating protein secondary structure prediction.

Table 10.1 summarizes neural network applications for protein structure prediction. Protein secondary structure prediction is often used as the first step toward understanding and predicting tertiary structure because secondary structure elements constitute the building blocks of the folding units. An estimated 90% or so of the residues in most proteins are involved in three classes of secondary structures, the a-helices, p-strands or reverse turns. Related to the secondary structure prediction are also the prediction of solvent accessibility, transmembrane helices, and secondary structure content (10.2). Neural networks have also been applied to protein tertiary structure prediction, such as the prediction of the backbones or side-chain packing, and to structural class prediction (10.3). 

Figure 10.1 Neural network for protein secondary structure prediction. (Adopted from Qian Sejnowski, 1988).

Protein secondary structure prediction is one of the earliest neural network applications in molecular biology, and has been extensively reviewed. Typified by the work of Qian and Sejnowski (1988) (Figure 10.1), early studies involved the use of perception or three-... 

Neural network method is often quoted as a data-driven method. The weights are adjusted on the basis of data. In other words, neural networks learn from training examples and can generalize beyond the training data. Therefore, neural networks are often applied to domains where one has little or incomplete understanding of the problem to be solved, but where training data is readily available. Protein secondary structure prediction is one such example. Numerous rules and statistics have been accumulated for protein secondary structure prediction over the last two decades. Nevertheless, these... 

Yao XQ, Zhu H, She ZS. A dynamic Bayesian network approach to protein secondary structure prediction. BMC Bioinformatics. 2008 9 49. 

An Evaluation Of Protein Secondary Structure Prediction Algorithms... 

Figure 3. Multidimensional scaling analysis of the dissimilarities between accuracies of different protein secondary structure prediction methods. The method codes can be found in Table I.

The present analysis might give rise to a somewhat pessimistic view of the effectiveness of protein secondary structure prediction algorithms. In fact, with the increasing number of proteins with known three-dimensional structure, constant re-evaluation of performance must take place in order to ascertain the validity of the methods. We note that the methods do not have the predictive power claimed by its authors when analyzed consistently using the 148 proteins selected in this study. Moreover, the situation is even worse for the Mathews correlation coefficient, which indicates that the predictions are poorly correlated with the actual structure. 

Leng, B., Bnchanan, B.G., and Nicholas, H.B., Protein Secondary Structure Prediction Using Two-Level Case-Based Reasoning, J. Comput. Biol., 1(1), 25, 1994. 

Frishman, D. and Argos, R, Incorporation of non-local interactions in protein secondary structure prediction from the amino acid sequence, Prot. Eng., 9, 133, 1996. 

---