Neural Network Prediction of Protein Structures

The main theme of this book is the prediction of protein structure using „artificial intelligence“ and, in particular, neural network methods. The first part of the text gives basic introductions both to protein structure and to neural networks. Part II then discusses the essential steps in implementing neural networks to predict, among other things protein secondary structures, protein fold classes, backbone and side-chain structures, and ligand binding sites. Protein contact dynamics is addressed in the final part. From the work presented here, it emerges that there is also a deeper philosophical connection between the optimization tasks performed by artificial neural networks and those that lead to observed protein structures.
TOC: 1. Introduction, 2. The basic structural elements of proteins, 3. Basics of neural networks, 4. Structural classification of folded proteins, 5. A statistical mechanical model for formation of protein fold classes, 6. Prediction of secondary structures by neural networks, 7. Neural networks prediction of protein fold classes, 8. Prediction of three-dimensional backbone structure of proteins, 9. Side-chain orientations in proteins predicted by Potts neural networks, 10. Prediction of ligand binding sites in proteins, 11. Surface prediction by neural networks, 12. Predicting the CDR-domains of antibodies by neural networks, 13. Formation of fold patterns in the early stages of protein folding, 14. Protein structure and chemical reaction kinetics, 15. Density funcional calculations for spectrosopcy in chemistry, 16. Neural network analysis of spectroscopic and structural calculations, 17. Future outlook