Fine-Tuning a Transformer Model for METTL3 Lead Optimization
Open Access
- 20 January 2026
- journal article
- Published by American Chemical Society (ACS) in ACS Bio & Med Chem Au
- Vol. 6 (2) , 160-167
Abstract
Transformers are machine learning models originally developed to translate between natural languages. Recently, a transformer model was trained on knowledge of medicinal chemistry, i.e., matched molecular pairs of nearly a million bioactive compounds from the ChEMBL database. Here, we customize (i.e., fine-tune) the pretrained model to enhance the affinity and/or metabolic stability of a series of inhibitors of methyltransferase-like protein 3 (METTL3). We first fine-tune the transformer model using a data set of about 500 METTL3 inhibitors with known binding affinities and validate it by retrospective analysis. Then, we fine-tune the original transformer model to simultaneously optimize binding affinity and metabolic stability in a prospective application. Two of the five METTL3 inhibitors predicted by the multiobjective optimized model show low-nanomolar potency and higher stability than the lead compound of the chemical series used for fine-tuning.Keywords
Funding Information
- Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (212195)
- Swiss Cancer Research Foundation (KFS-5748-02-2023)
- Lotte und Adolf Hotz-Sprenger Stiftung
This publication has 37 references indexed in Scilit:
- Chemical language models for de novo drug design: Challenges and opportunitiesCurrent Opinion in Structural Biology, 2023
- Machine Learning in Drug Discovery: A ReviewArtificial Intelligence Review, 2021
- Assessment of the Fragment Docking Program SEEDJournal of Chemical Information and Modeling, 2020
- A Reader-Based Assay for m6A Writers and ErasersAnalytical Chemistry, 2019
- The m 6 A Methyltransferase METTL3 Promotes Translation in Human Cancer CellsMolecular Cell, 2016
- Nuclear m6A Reader YTHDC1 Regulates mRNA SplicingMolecular Cell, 2016
- N6-methyladenosine Modulates Messenger RNA Translation EfficiencyCell, 2015
- Efficient electrostatic solvation model for protein-fragment dockingProteins: Structure, Function, and Bioinformatics, 2000
- Exhaustive docking of molecular fragments with electrostatic solvationProteins: Structure, Function, and Bioinformatics, 1999
- Methylated nucleotides block 5′ terminus of HeLa cell messenger RNACell, 1975