What I work on
Machine learning and language models for drug discovery and toxicology.
Under review
Chemical Language Models for Early-Stage Drug Discovery: Applications, Pitfalls, and Future Directions
A review of what chemical language models are used for in early drug discovery and where they break down.
Computational Flagging of Assay Interference in Drug Discovery: A Review of Rule-Based, Machine Learning, Graph Neural Networks, and Chemical Language Models
A review of the methods used to flag assay interference compounds, from substructure rules through to graph neural networks and chemical language models.
Published
Computational approaches to adverse outcome pathway networks
A review of how adverse outcome pathway networks are built, quantified and used to support regulatory decisions.
GitHub in AI-driven drug discovery
How sharing code on GitHub makes AI drug discovery research easier to check and reproduce.
Predicting drug-drug interactions with AutoML
Automated machine learning for drug-drug interactions, and how the way a molecule is represented changes the result.
Language models in drug delivery
A review of language model tools, such as ChemCrow and BioBERT, for formulation and drug delivery design.
How reliable are ML predictions of tumor progression?
A close look at how machine learning models of tumor progression are benchmarked, where dataset bias creeps in, and what reliable should mean.
BAD Molecule Filter
My MSc work. A model that flags molecules that look like real hits in a drug screen but are actually clumping together and causing false positives. It runs as a free public web server.
Now
AI agents for drug discovery
Language model agents that plan a drug discovery task and call real chemistry tools to carry it out, without a person steering every step.
Looking for: A collaborator who builds LLM agents, or who knows target discovery.
AI for adverse outcome pathways
Using graph learning and language models to build Adverse Outcome Pathway networks, the maps that link a molecular event to a toxic effect.
Looking for: Toxicologists, graph ML researchers, or teams doing regulatory toxicology.
Working on something close to this? Email me.