Tess: A geometric hashing algorithm for deriving 3D coordinate templates for searching structural databases. Application to enzyme active sites

Abstract

It is well established that sequence templates such as those in the PROSITE and PRINTS databases are powerful tools for predicting the biological function and tertiary structure for newly derived protein sequences. The number of X-ray and NMR protein structures is increasing rapidly and it is apparent that a 3D equivalent of the sequence templates is needed. Here, we describe an algorithm called TESS that automatically derives 3D templates from structures deposited in the Brookhaven Protein Data Bank. While a new sequence can be searched for sequence patterns, a new structure can be scanned against these 3D templates to identify functional sites. As examples, 3D templates are derived for enzymes with an 0-His-0 "catalytic triad' and for the ribonucleases and lysozymes. When these 3D templates are applied to a large data set of nonidentical proteins, several interesting hits are located. This suggests that the development of a 3D template database may help to identify the function of new protein structures, if unknown, as well as to design proteins with specific functions.

Using Bregmann Divergence Regularized Machine for Comparison of Molecular Local Structures

“…In the Local Structure Search stage, a geometrical hashing algorithm such as TESS [4] or JESS [5] is used to enumerate local structures whose shapes are possibly similar to the model structure. The local structures contain every atom in the model structure.…”

SEGA: Semiglobal Graph Alignment for Structure-Based Protein Comparison

“…Methods in this field employ geometric hashing [21,22], dynamic programming [23], graph theory [24,7] and other strategies [25,26]. Most of these approaches scan user defined or automatically generated templates against a database to detect frequent patterns.…”

Structural Descriptors of Protein-Protein Binding Sites

“…In contrast to the data-driven approach by Henschel et al 11 outlined above, this procedure requires less data, but relies heavily on the structural alignment method, making it difficult to differentiate between differences in binding sites and methodological artefacts. For the comparison of enzymatic active sites the software packages TESS 32 and JESS 3 were developed by the Thornton group. The Klebe group developed clique search and geometric hashing approaches for the comparison of small-ligand binding sites.…”