Novel Integrated Approach to Structuring Databases of Secondary and 3-D Protein Structures: Application to Immunoglobulin Structure Analysis and Prediction

Israel Gelfand, Alexander Kister

A highly novel approach based on the extraction of significant features of protein structure has resulted in a striking finding: the derivation of an invariant system of coordinates that permits comparison both within and between the data characterizing variable domains in the proteins. These results have followed from the analysis of 5000 Ig sequences (from the Kabat database) from which the secondary structures were acurately predicted. The initial structural analysis of the Ig database has been published in the Proceedings of the National Academy (Gelfand and Kister, November 1995), and the results for the invariant coordinate system has been submitted for publication in December 1995.

We are proposing to extend the analytical approach we have developed to systematize it as a means of determining relations between sequence, secondary, and three-dimensional structures for proteins in general.

The major elements of the novel approach are as follows:

A new database organization of secondary structures for Ig molecules was created, where every residue was assigned a position in a strand or loop and a characteristic position derived.
A statistical analysis of the residues was developed based on a novel decomposition in terms of structural units.
Examining 60 three dimensional crystalographic structures of Ig molecules permited us to define the different structural roles of the residues in the positions for folding of the Ig chain. Using this classification we identified the position in the database [PNAS November paper].

Subsequent to the results noted above we investigated a new approach for comparing protein structures, since the above did not yield an optimal superposition of structure based on inherent geometrical properties of the molecules.

The result of our most recent research, as mentioned at the begining of this proposal, is an invariant system of coordinates for Ig molecules. Comparison of 3 D structures of the mouse kappa molecules allows us to find the common features of protein folding and to determine the conservative C-alpha framework for the proteins.

When generalized to other protein structures, the implications of this novel method of integrated analysis can be extremely significant. It will allow correlations between sequence, secondary and 3-D strucutre which do not appear obvious otherwise, and can lead to important insights into protein function, opening many new possibilities for therapy and drug design.