Review of the SciProtein Molecular Modeling Program1

Since the seminal work of C. Anfinsen, it has been accepted that the primary amino acid sequence of a protein directs its folding into secondary and tertiary functional form. Though elegant in its simplicity, the understanding of how this primary sequence actually directs the three-dimensional folding of a complicated structure, with literally an astronomical number of available conformations, remains a substantial practical problem and one of fundamental interest to theoretical computational chemists. With the growing knowledge of primary sequence, mostly from the success of the human genome project, this question of discovering a protein folding code that is linked to the pattern of DNA/ RNA protein production is an important area of active research. However, most biological scientist are unwilling to wait for the full deciphering of this physical/chemical encryption scheme. To this end, numerous techniques have been developed in computational chemistry laboratories to assist at least with a partial prediction of structure based on primary code. There is an ever-increasing array of commercial products designed to assist with the manipulation and exploration of primary sequence; most of these relate primary and secondary sequence structure. SciProtein, by SciVision (Lexington, MA), has recently brought out Version 2 of their program that engages this problem. SciProtein is described as a " comprehensive tool kit for proteins and peptides " and is an integrated package structured around homology search and structural prediction engines. The homology engine is capable of searching the Brookhaven database using the FASTP Algorithm and allows search of both primary and secondary structure finding elements or complete proteins with similar primary and secondary structure to a peptide under investigation. The prediction engine is based on the work of R. J. Gilbert and uses a prediction scheme utilizing a digital encoding algorithm in which a set of training proteins is used to generate a database. Based on the principle that the secondary structure is influenced by the local already formed secondary structural domains, the database is drawn from a set of known protein structures related to the function or milieu of the unknown protein. Thus secondary structure prediction proceeds based on standard statistical methods (hydropathy, etc.) that are modified and guided in a heuristic manner by the empirical database. Using properly chosen training sets, the prediction scheme can be somewhat superior to the more commonly used statistical methodologies. SciProtein is integrated with two of the more popular PC-based programs, HyperChem and …