No abstract

The existing literature that models item-by-item production yield as a Bernoulli process assumes that the intraperiod likelihood of producing an acceptable item is stationary. We investigate the stochastic process that results from relaxing this assumption to account for system deterioration during each production run. More specifically, we consider a Bernoulli yield model with a nonstationary parameter that depends on the deterioration level of the system, which evolves according to a discrete-time Markov chain. For tractability reasons, we construct a simple binomial approximation of the non-stationary process, and compare the two yield distributions both analytically and numerically. Our results suggest that the approximation performs well, even when the deterioration occurs relatively fast, which serves to validate existing (and future) decision models that impose the stationarity assumption.

AI/ML methods in drug discovery are maturing and their utility and impact is likely to permeate many aspects of drug discovery including lead finding and lead optimization. Typical methods utilize ML-models for structure-property prediction with simple 2D-based chemical representations of the small molecules. Further, limited data, especially pertaining to novel targets, make it difficult to build effective structure-activity ML-models. Here we describe our recent work using the BIOVIA Generative Therapeutics Design (GTD) application, which is equipped to take advantage of 3D structural models of ligand protein interaction, i.e., pharmacophoric representation of desired features. Using an SAR data set pertaining to the discovery of SYK inhibitors entospletinib and lanraplenib in addition to two unrelated clinical SYK inhibitors, we show how several common problems in lead finding and lead optimization can be effectively addressed with GTD. This includes an effort to retrospectively re-identify drug candidate molecules based on data from an intermediate stage of the project using chemical space constraints and the application of evolutionary pressure within GTD. Additionally, studies of how the GTD platform can be configured to generate molecules incorporating features from multiple unrelated molecule series show how the GTD methods apply AI/ML to drug discovery.

Drug discovery requires the simultaneous optimization of many properties such as bioactivity, absorption, distribution, metabolism, excretion, toxicity, and the underlying physicochemical properties. The ability to consider many such requirements is termed multiobjective optimization, and we will discuss the importance of this for drug discovery with a specific focus on ADME/Tox. In particular, we provide examples of Pareto optimization methods and how they can be used in drug discovery to make trade‐offs between these different predicted or real molecular properties and describe a computer program, Pareto Ligand Designer, created using Pipeline Pilot™.

Drug discovery requires the simultaneous optimization of many properties such as bioactivity, absorption, distribution, metabolism, excretion, toxicity, and the underlying physicochemical properties. The ability to satisfy many requirements at once is termed multiobjective optimization, and we will discuss the importance of research in this area for drug discovery and development. In particular, we provide examples of Pareto and other optimization methods and how they can be used in drug discovery to make trade‐offs between different predicted or real molecular properties, and we describe advances in software that applies these approaches.