Density functional theory and Bayesian optimization for therapeutic discovery

Authors: Kaijage, Nadhiri; Sewunet, Sosina; Otieno, Edward; Kraszewski, Sebastian

Source journal: Expert opinion on drug discovery (2026-11-16)

DOI: 10.1080/17460441.2026.2762802

Abstract

INTRODUCTION: The rational design of novel therapeutics is fundamentally constrained by an irreducible paradox: the drug-like chemical space spans an estimated 10(60)-10(63) compounds, yet quantum mechanical evaluation of molecular electronic structure demands methods scaling as O(N(3))-O(N(4)) with system size. Bayesian Optimization-guided Density Functional Theory resolves this by embedding quantum-accurate property evaluation within a probabilistic active learning loop that directs computational resources toward maximally informative candidates. AREAS COVERED: The authors critically examine DFT functional selection, basis set adequacy, and scaling mitigation strategies for drug-like molecules, alongside Bayesian Optimization surrogate architectures, acquisition function design, and multi-fidelity extensions spanning GFN2-xTB to DLPNO-CCSD(T). A quantitative structured comparative synthesis of eleven published DFT-BO campaigns is presented, and integration strategies, including sequential, batch, and surrogate-assisted pipelines, are analyzed. This review is based on a structured literature search conducted in PubMed, Web of Science, Scopus, and Google Scholar, covering publications from 1965 to July 2026. EXPERT OPINION: The convergence of fault-tolerant quantum computing, large-scale molecular foundation models, and exascale hardware is expected to broaden the scope of DFT-BO integration. We argue that the value of these methods is presently established for the optimization of quantum-chemical properties rather than for therapeutic endpoints, and we identify the benchmarking and validation evidence that would be required to establish the latter.

Read the full text

Related topics

Related open access articles

Abstract and bibliographic data from PubMed (U.S. National Library of Medicine). Copyright of the abstract belongs to the original publisher or authors.