Modelling iterative compound optimisation using a self-avoiding walk

Abstract

The optimisation phase is a crucial step in the process of drug development, yet the mechanics of the projects that make it up are poorly understood. Weak documentation of failed projects makes statistical analysis of the factors affecting project performance challenging, so a better approach may be the development of an underlying theory of how projects work. We present a model based on a modified random walk in a relevant chemical space and use it to produce simulations of projects and portfolios of projects. Simulation is used to explore parameters that might affect the performance of a project and shows that they fall into two groups--target and process--that affect the overall performance in distinct ways.