Molecular modeling of protein structure and function: a bioinformatic approach.

This paper reports on the data/information structure of macromolecules as it extends beyond the three-dimensional conformation to include functional descriptors of biochemical (in vitro) and biological (in vivo) characteristics and as it contrasts with the limitations imposed by the data reduction and data classification techniques of traditional molecular modeling. Methodologies for structure-function representation are presented which are being incorporated within a knowledge-acquisition expert system. Examples of the bioinformatic approach are presented concerning macromolecular recognition by serine proteases and the use of Fourier transform-infrared (FT-IR) spectroscopy for structural assignment and analysis by a novel structure-perturbation approach.