Machine learning methods for pKa prediction of small molecules: Advances and challenges. Issue 12 (December 2022)

• Record Type: Journal Article
• Authors: Wu, Jialu; Kang, Yu; Pan, Peichen; Hou, Tingjun
• Is Part Of: Drug discovery today. Volume 27:Issue 12(2022)
• Publication Date: 2022-12
• Subjects: pKa prediction -- QSAR -- Machine learning -- Handcrafted features -- Graph neural networks; Drugs -- Design -- Periodicals; Drugs -- Research -- Periodicals; 615.1
• DOI: 10.1016/j.drudis.2022.103372
• Languages: English
• ISSNs: 1359-6446

Abstract

Highlights: QSAR models for p K a prediction comprise descriptor-based and graph-based approaches. Data scarcity and the intrinsic complexity of p K a are major challenges for prediction. Graph neural networks combined with domain knowledge are powerful and promising. p K a prediction tools will hopefully be a significant component in AI-driven drug design. Abstract : The acid–base dissociation constant (p K a ) is a fundamental property influencing many ADMET properties of small molecules. However, rapid and accurate p K a prediction remains a great challenge. In this review, we outline the current advances in machine-learning-based QSAR models for p K a prediction, including descriptor-based and graph-based approaches, and summarize their pros and cons. Moreover, we highlight the current challenges and future directions regarding experimental data, crucial factors influencing p K a and in silico prediction tools. We hope that this review can provide a practical guidance for the follow-up studies.