Exploration and Exploitation Approaches Based on Generative Machine Learning to Identify Potent Small Molecule Inhibitors of α‑Synuclein Secondary Nucleation

Aug 6, 2026

Upload by Logic

High attrition in drug discovery pipelines poses a major challenge for Parkinson’s disease, where no disease-modifying therapies have been approved and many α-synuclein aggregation–targeting trials have failed. The work introduces a machine learning framework combining generative modeling and reinforcement learning to discover small molecules that reshape aggregation kinetics by reducing oligomer formation. Training uses assay data quantifying inhibition of α-synuclein secondary nucleation, the key mechanism for oligomer production, yielding candidates with strong secondary-nucleation potency.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Targeting α-synuclein oligomer formation and secondary nucleation

Generative modeling and reinforcement learning for inhibitor discovery

QSAR-guided filtering to reduce candidate testing

[pubs.acs.org/JCTC](pubs.acs.org/JCTC) Article Exploration and Exploitation Approaches Based on Generative Machine Learning to Identify Potent Small Molecule Inhibitors of α‑Synuclein Secondary Nucleation Robert I. Horne,\# Mhd Hussein Murtada,\# Donghui Huo,\# Z. Faidon Brotzakis, Rebecca C. Gregory, Andrea Possenti, Sean Chia, and Michele Vendruscolo * Cite This: J. Chem. Theory Comput. 2023, 19, 4701−4710 Read Online | | | | | | | | --- | --- | --- | --- | --- | --- | | ACCESS | Metrics & More | | Article Recommendations | | *sı Supporting Information | ABSTRACT: The high attrition rate in drug discovery pipelines is an especially pressing issue for Parkinson’s disease, for which no disease-modifying drugs have yet been approved. Numerous clinical trials targeting α-synuclein aggregation have failed, at least in part due to the challenges in identifying potent compounds in preclinical investigations. To address this problem, we present a machine learning approach that combines generative modeling and reinforcement learning to identify small molecules that perturb the kinetics of aggregation in a manner that reduces the production of oligomeric species. Training data were obtained by an assay reporting on the degree of inhibition of secondary nucleation, which is the most important mechanism of α -synuclein oligomer production. This approach resulted in the identification of small molecules with high potency against secondary nucleation. ■ INTRODUCTION A link between α-synuclein (αS) aggregation and Parkinson’s disease (PD) is supported by genetic evidence and by observations of the accumulation of αS in inclusions known as Lewy bodies in the brains of PD patients. 1−3 A primary aim of current research toward the development of therapeutic treatments of this disease is therefore the inhibition of αS aggregate formation. Our approach here is based on the realization that it is particularly important to target αS aggregation by specifically preventing the formation of αS oligomers.4 These intermediate species are particularly cytotoxic, as they can disrupt cell membranes, especially those of mitochondria.5−7 In order to reduce the number of oligomers produced in an aggregation reaction, one should take into account that highly ordered fibrillar aggregates can act as highly effective catalytic surfaces for oligomer formation.8 The pathological relevance of these processes has led to major investment into identifying compounds that can inhibit those aggregation mechanisms associated with neurotoxicity.9−12 As therapies are beginning to be delivered for Alzheimer’s disease,13 the race is on to achieve the same outcome for PD patients. 14−16 Computational methods can contribute to these endeavors. In particular, in recent years, deep learning has emerged as a powerful tool for cheminformatics.17 With this capability, molecular generative models have emerged as promising tools for de novo molecular design. It has also been previously shown that computational methods can offer more efficient routes to αS aggregation inhibitors than traditional screening methods.18, 19 A limitation of that approach was the use of preexisting screening libraries, which biased the model and limited the search space. A further limitation was focusing only on the molecule potency during the machine learning task. The present work aims at addressing these shortcomings through the application of generative modeling approaches and multiparameter optimization in two separate pipelines: (1) one focused on exploration (identifying novel and effective molecular structures), and (2) the other on exploitation (achieving higher potency from known chemical space). The former employs an architecture derived from the GraphINVENT20 framework for multiparameter generative modeling while the latter consists of a chemical language model optimized for low data regimes.21 Both pipelines feature a generative model linked to a QSAR (qu

FAQs

Why is secondary nucleation targeted for α-synuclein inhibitor discovery?

Secondary nucleation is presented as the most important mechanism driving oligomer production. Reducing secondary nucleation lowers oligomeric species associated with toxicity.

What machine learning components does the approach combine?

The method combines generative modeling with reinforcement learning, using a generative model coupled to QSAR-based activity filtering to guide molecule generation.

How were training data and scoring functions obtained for the models?

Training relies on assay measurements that report inhibition of secondary nucleation. Scoring functions are based on QSAR molecular activity classifiers trained on experimental data.

Aug 6, 2026

Upload by Logic