Machine learning methods for pKa prediction of small molecules: Advances and challenges

•QSAR models for pKa prediction comprise descriptor-based and graph-based approaches.•Data scarcity and the intrinsic complexity of pKa are major challenges for prediction.•Graph neural networks combined with domain knowledge are powerful and promising.•pKa prediction tools will hopefully be a signi...

Full description

Saved in:
Bibliographic Details
Published in:Drug discovery today 2022-12, Vol.27 (12), p.103372-103372, Article 103372
Main Authors: Wu, Jialu, Kang, Yu, Pan, Peichen, Hou, Tingjun
Format: Article
Language:English
Subjects:
Graph neural networks
Handcrafted features
Machine learning
pKa prediction
QSAR
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•QSAR models for pKa prediction comprise descriptor-based and graph-based approaches.•Data scarcity and the intrinsic complexity of pKa are major challenges for prediction.•Graph neural networks combined with domain knowledge are powerful and promising.•pKa prediction tools will hopefully be a significant component in AI-driven drug design. The acid–base dissociation constant (pKa) is a fundamental property influencing many ADMET properties of small molecules. However, rapid and accurate pKa prediction remains a great challenge. In this review, we outline the current advances in machine-learning-based QSAR models for pKa 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 pKa and in silico prediction tools. We hope that this review can provide a practical guidance for the follow-up studies.
ISSN:1359-6446
1878-5832
DOI:10.1016/j.drudis.2022.103372