Predicting Protein Tertiary Structure

We began the discussion of globular protein tertiary structure by pointing out that the secondary and tertiary structure is determined by the primary structure and that this is probably a reflection of the fact that the native folded conformation is the most stable structure that can be formed. If this is so, then it should be possible to predict a protein s structure from its primary sequence. At this juncture, such predictions remain an elusive goal. However, most proteins are made of a limited number of domains, which tend to reappear in many different proteins. Since this is the case, it may be possible to predict the structures of many proteins in the future by using the information accumulated from x-ray diffraction studies of related proteins. 

It is clear that certain amino acids tend to form particular secondary structures. As shown in figure 4.23 glutamic 

Relative probabilities that any given amino acid occurs in the a-helical, /3-sheet, or /3-hairpin-bend secondary structural conformations. 

Quaternary Structure Involves the Interaction of Two or More Proteins 

A second, commonly observed pattern of quaternary structure is for a molecular aggregate to have multiple cop- 

---

Visualizing Folded Protein Structures Primary Structure Determines Tertiary Structure Secondary Valence Forces Are the Glue That Holds Polypeptide Chains Together Domains Are Functional Units of Tertiary Structure Predicting Protein Tertiary Structure Quaternary Structure Involves the Interaction of Two or More Proteins... 

The most general approach attempts to predict protein tertiary structure from sequence without any recourse to known protein structures or evolutionary information. This is the traditional approach to the solution of the protein... 

One of the common ways of predicting protein tertiary structure assumes that the secondary structure has to be known before the prediction of a three-dimensional fold can be attempted [154-157]. While this view as a paradigm for protein structure prediction could be challenged, it certainly provides a straightforward framework that may sometimes prove to be useful. Indeed, there were a number of early attempts to apply such a methodology to low resolution protein fold predictions that were quite successful in some specific cases [154-157]. However, only recently has the problem of protein structure assembly, given its secondary structure, been more systematically addressed. 

S. Sun, Reduced representation approach to protein tertiary structure prediction statistical potential and simulated annealing, J. Theor. Biol. 172 (1995), 13-32. 

Galaktionov, S.G., Marshall, G.R. Properties of intraglobular contacts in proteins an approach to prediction of tertiary structure. Proceedings of the 27th Hawaii International Conference on System Sciences. IEEE Computational Society, Washington, DC, 1994,... 

H. Kawai, T. Kikuchi, and Y. Okamoto, Protein Engin., 3, 85 (1989). A Prediction of Tertiary Structures of Peptide by the Monte Carlo Simulated Annealing Method. 

Le-Grand SM, Merz KM Jr (1994) The protein folding problem and tertiary structure prediction the genetic algorithm and protein tertiary structure prediction. Birkhauser, Boston, p 109... 

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). 

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. 

The earliest neural network attempt for protein tertiary structure prediction was done by Bohr et al. (1990). They predicted the binary distance constraints for the C-a atoms in protein backbone using a standard three-layer back-propagation network and BIN20 sequence encoding method for 61-amino acid windows. The output layer had 33 units, three for the 3-state secondary structure prediction, and the remaining to measure the distance constraints between the central amino acid and the 30 preceding residues. 

The basic information of protein tertiary structural class can help improve the accuracy of secondary structure prediction (Kneller et al., 1990). Chandonia and Karplus (1995) showed that information obtained from a secondary structure prediction algorithm can be used to improve the accuracy for structural class prediction. The input layer had 26 units coded for the amino acid composition of the protein (20 units), the sequence length (1 unit), and characteristics of the protein (5 units) predicted by a separate secondary structure neural network. The secondary structure characteristics include the predicted percent helix and sheet, the percentage of strong helix and sheet predictions, and the predicted number of alterations between helix and sheet. The output layer had four units, one for each of the tertiary super classes (all-a, all-p, a/p, and other). The inclusion of the single-sequence secondary structure predictions improved the class prediction for non-homologous proteins significantly by more than 11%, from a predictive accuracy of 62.3% to 73.9%. 

In this chapter, some methods currently employed in the prediction of tertiary structure from the protein sequence will be highlighted This is not intended to be a comprehensive review of the literature, a goal that would require an entire book of this size and would quickly become outdated. Instead, the aim is to provide some of the basic tenets of the theories and methods utilized with a few select references to the original literature. Hopefully this will serve as a starting point to delve into this very exciting and ongoing field of research. 

Aloy, R, and R. B. Russell. 2003. InterPreTS Protein interaction prediction through tertiary structure. Bioinformatics 19 161-2. 

Brocklehurst, S.M. and Perham, R.N. (1993) Prediction of the three-dimensional structures of the biotinylated domain from yeast pyruvate carboxylase and of the lipoylated H-protein from the pea leaf glycine cleavage system a new automated method for the prediction of protein tertiary structure. Protein Sci. 2 626-639. 

Can we predict the tertiary structure of a protein if we know its amino acid sequence ... 

Tan, C.-W., Jones, D. Using neural networks and evolutionary information in decoy discrimination for protein tertiary structure prediction. BMC Bioinformatics 2008, 9, 94. 

Zhang, Y, Arakaki, A.K., Skolnick, J. TASSER an automated method for the prediction of protein tertiary structures in CASP6, Proteins 2005, 61(Suppl. 7), 91-8. 

---