Applications - Protein Structure Prediction

Reference Application Neural Network Input/Output Encoding 

Rost Sander, 1993a b 2-D Structure 2x3L/FF/BP Prf21 Real3/3(a,PL-) 

Rost Sander, 1994a 2-D Structure 2x3L/FF/BP Prf21+Real3+%AA/3(a,p,L) Real4+Reall+%AA/3(a,P,L) 

Rost etal., 1995 Transmembrane Helices 2x3L/FF/BP Prf21+Real3+%AA/2(Y,N) Real3+Real l+% AA/2(Y,N) 

---

Fukunishi, O. Watanabe Takada, S., On the Hamiltonian replica exchange method for efficient sampling of biomolecular systems application to protein structure prediction, J. Chem. Phys. 2002,116, 9058-9067... 

In lattice models, the location of each element on the lattice can be stored as a vector of coordinates [(X, F,), (X2, Y2), (X3, Y3),..., (Xn, F )], where (X Y,) are the coordinates of element i on a two-dimensional lattice (a three-dimensional lattice will require three coordinates for each element). Since lattices enforce a fixed geometry on the conformations they contain, conformations can be encoded more efficiently by direction vectors leading from one atom (or element) to the next. For example in a two-dimensional square lattice, where every point has four neighbors, a conformation can be encoded simply by a set of numbers (Lu L2, L3,..., L ), where L, g 1, 2,3,4 represents movement to the next point by going up, down, left, or right. Most applications of GAs to protein structure prediction utilize one of these representations. 

Table 10.1 Neural network applications for protein 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). 

How have neural networks been used in genome informatics applications In Part II, we have summarized them based on the types of applications for DNA sequence analysis, protein structure prediction and protein sequence analysis. Indeed, the development of neural network applications over the years has resulted in many successful and widely used systems. Current state-of-the-art systems include those for gene recognition, secondary structure prediction, protein classification, signal peptide recognition, and peptide design, to name just a few. 

Some bioinformatics problems can be formulated as MOOPs, for instance the sequence alignment of DNA or protein sequences, protein structure prediction and design, and inference of gene regulatory networks, just to mention a few. The interested reader is referred to the review of Handl et al. (2007) that covers in detail more MOO applications in bioinformatics. The next few paragraphs describe some of the current applications in bioinformafics of interest to the chemical engineering community. 

This procedure is applied ubiquitously in protein structure prediction. Despite the caveats that the Boltzmann law is only applicable in thermal equilibrium and that it cannot readily be applied additively to partial molecular states scoring functions based on inverse Boltzmann statistics have proved very valuable in protein structure prediction. 

F, E. Cohen, L. Gregoret, S. R. Presnell, and I. D. Kuntz, in Computer-Assisted Modeling of Receptor-Ligand Interactions Theoretical Aspects and Applications to Drug Design. R, Rein and A. Golombek, Eds., Liss, New York, 1989, pp. 75-85. Protein Structure Predictions New Theoretical Approaches. 

One of the greatest challenges facing theorists today is the prediction of the 3-D structure of a protein starting from its amino acid sequence. Due to the size and complexity of proteins, theoretical methods currently used to investigate small organic molecules are not directly applicable in the study of protein structure. Current approaches to protein-structure prediction can be broadly classified into a priori and heuristic methods. A number of reviews [1-5] and books [6,7] which address 3-D-structure prediction and theoretical aspects of protein folding are available and should be consulted for additional details. 

One of the first and probably the most prominent example for such an application was the attempt to predict the secondary structure of proteins by means of an error-backpropagation ANN (Figure 13). Windows of 13 amino acid sequences taken from various proteins were used as input. On the output (taiget) side three neurons were employed to distinguish between one of the three possible states of the secondary structure in which the amino acid at the middle of the window sequence could be either a-helix, -sheet, or coil (see Protein Modeling and Protein Structure Prediction in ID, 2D, and 3D). Because 20 different amino acids plus a terminator were encoded with... 

Artificial Intelligence in Chemistry Chemical Engineering Expert Systems Chemometrics Multivariate View on Chemical Problems Electrostatic Potentials Chemical Applications Environmental Chemistry QSAR Experimental Data Evaluation and Quality Control Fuzzy Methods in Chemistry Infrared Data Correlations with Chemical Structure Infrared Spectra Interpretation by the Characteristic Frequency Approach Machine Learning Techniques in Chemistry NMR Data Correlation with Chemical Structure Protein Modeling Protein Structure Prediction in ID, 2D, and 3D Quality Control, Data Analysis Quantitative Structure-Activity Relationships in Drug Design Quantitative Structure-Property Relationships (QSPR) Shape Analysis Spectroscopic Databases Structure Determination by Computer-based Spectrum Interpretation. 

---