2D-Structure Drawings of Similar Molecules

Abstract

A common strategy in drug design and pharmacophore identification consists of evaluating large sets of molecular structures by comparing their 2D structure drawings. To simplify the chemists' task, the drawings should reveal similarities and difierences between drugs. Given a family of molecules all containing a common template, we present an algorithm to compute standardised 2D structure drawings. The molecules being represented as a graph, we compute a structure called supertree in which all molecules can be embedded. Using the correspondences between atoms provided by the supertree, we are able to coordinate the drawings performed by a breadth-first traversal of the molecular graphs. Both parts of the problem are NP-hard. We propose algorithms of heuristic nature.