Knowledge-based modeling

It has long been recognized that proteins with similar amino acid sequences have similar three-dimensional protein structures. Differences between homologous structures usually occur in loops on the protein surface. The framework or core residues are conserved owing to constraints on the composition of interior side chains and other structural restraints. When a sequence is homologous to a protein of known structure, a knowledge-based approach to modeling is a very 

Protein segments adopt only a finite number of conformations in folded proteins.This limited dictionary of templates may be due to the inherent steric constraints of the backbone and other packing and electrostatic interactions in protein folding. This was most graphically demonstrated by Jones and Thirup. They showed that the majority of the polypeptide backbone of a new structure can be built up from component pieces of other structures, whether or not they are related. Unger et al. showed, in fact, that 76% of the main chain structure of an 82 protein data set can be constructed using only 100 unique hexapeptide templates. Amino acid side chains are also found in a restricted set of conformers. Recent work in our laboratory has shown that a set of four tetrapeptide templates is sufficient to define almost all loop conformations. 

Cohen and co-workers considered rigid units of secondary structure and combinatorially generated all possible three-dimensional arrangements between them. Chain connectivity, topological restrictions, and steric constraints were applied to eliminate incorrect or impossible structures. The number of resultant structures was reduced to a manageable set (on the order of 10-100), and the correct structure was always present within the set of solutions. There was no way to differentiate the correct answer from the other possibilities, however. Indeed, Novotny and coworkers showed that it is difficult to correctly 

---

J Bajorath, R Stenkamp, A Amffo. Knowledge-based model building of proteins Concepts and examples. Protein Sci 2 1798-1810, 1994. 

MJ Sutcliffe, I Haneef, D Carney, TL Blundell. Knowledge based modelling of homologous proteins. Part I Three dimensional frameworks derived from the simultaneous superposition of multiple structures. Protein Eng 1 377-384, 1987. 

MJ Sutcliffe, ERE Hayes, TL Blundell. Knowledge based modeling of homologous proteins. Part II Rules for the conformation of substituted side-chains. Protein Eng 1 385-392, 1987. 

J. Klaessens, T. Saris, B. Vandeginste and G. Kateman, Expert system for knowledge-based modelling of analytical laboratories as a tool for laboratory management. J. Chemometrics, 2 (1988)49-65. 

The following list divides types of unsafe acts into those that are skill-hased, perceptual, and decision or knowledge-based. (Note that this division is different from the skills-rules-model knowledge-based model described earlier.) Examples are given of each type of error that may result in an unsafe act.( )... 

C. Vinals, X. De Bolle, E. Depiereux, and E. Feytmans, Knowledge-based modeling of the D-lactate dehydrogenase three-dimensional structure, Proteins 1995, 21, 307-318. 

Shah PB, Kokossis AC. Knowledge-based models for the analysis of complex separation processes. Comp Chem Eng 2001 25 867. 

Knowledge-based model combining sequence data to other information, such as homology modeling (Hilbert et ah, 1993 Chinea et al., 1995). 

Devulapalle KS, Goodman SD, Gao Q, Hemsley A, Mooser G (1997) Knowledge-based model of a glucosyltransferase from the oral bacterial group of mutans streptococci. Protein Sci 6 2489-2493... 

Sutcliffe, M. J., Hayes, F. R. F., Blundell, T. L. Knowledge-based modelling of homologous proteins, part II Rules for the conformations of substituted sidechains. Protein Engineering, 1987, 1,385-392. 

As the subtitle suggests, the field can be viewed from two angles that of physical modelling and that of knowledge-based modelling. While the former is concerned with the physical interactions of matter from the electronic all the way to the macroscopic level, the latter deals with statistical relationships between models of structure and property data. The emerging themes in each of these areas seem at first sight disparate. 

In contrast, the emerging theme in knowledge based modelling is an ever closer intertwining of modelling and experimentation. A powerful combination of three factors contribute to this theme (i) structure-property relationships that are based on better models and more relevant structural descriptors (ii) robotics techniques in experimentation, which lead to a real explosion in the amount of data available (iii) informatics tools, which can handle the large amounts of data generated by both experiment and simulation. 

Knowledge-based Modelling QSPR/QSAR Methods and Neural Networks. 

Coming back to the issue raised in the introduction physical modelling to replace experiment and knowledge-based modelling to couple experiment and simulation more closely the two opposing trends My conclusion is that they are in fact complementary improvements in the physical modelUng will lead to better models and descriptors for correlation based approaches, while more... 

VeDa is intended as a modeling language for the domain of process engineering, which can be used for the representation of specific mathematical models of chemical process systems in a model library as well as for the representation of knowledge about models and the modeling process in a knowledge-based modeling environment. 

M. J. Sutcliffe, I. Haneef, D. Carney, and T. L. Blundell, Protein Eng., 1, 377 (1987). Knowledge-Based Modelling of Homologous Proteins. Part I. Three-Dimensional Frameworks Derived from the Simultaneous Superposition of Multiple Structures. 

---