Topomer selection among all accessible structures

Period of Performance: 04/01/2003 - 12/31/2003


Phase 1 SBIR

Recipient Firm

Tripos, Inc.
Saint Louis, MO 63144
Principal Investigator


DESCRIPTION (provided by applicant): The objective is a general methodology for using knowledge of one or more active ligands to select from among the vast realm of all synthetically accessible structures the most promising compounds for some particular therapeutic objective. A novel topomeric descriptor has been providing very rapid shape similarity searching of combinatorial "virtual libraries." Structures so identified but not otherwise apparently similar to a query structure have repeatedly been found upon synthesis to share a query's activity, unusually often. Recent extensions of the topomer methodology include: pharmacophoric features as well as shape; the searching of familiar structurally heterogeneous structures as well as virtual libraries; and 3D-QSAR-based as well as similarity searching, thereby enabling topomer searching for lead optimization as well as discovery. However the current technology would not succeed in "discovering" structures whose syntheses involve many steps and/or synthons that are not commercially available, including most drugs on the market. We therefore propose to develop an "iterative" version of topomer searching, which would search for matching structures one synthetic step at a time, guided by the constraints of earlier (retrosynthetically) synthetic steps as well as the overall query shape. Its starting synthons would include "custom-built synthons," those embedded in pharmacologically important structures, as well as those commercially available. In Phase I, we would implement several underlying algorithms. We would assess efficacy by demonstrating that a random set of pharmacologically interesting query structures will discover their own syntheses by a topomeric rationale (noting that the most similar structure to a query is the query structure itself), as well as other structures likely to be biologically similar. We would also assess how search speed and output scale with database size.