Drug Repurposing using Artificial Intelligence

An unwelcome cyclone may have cancelled my hiking plans, but it gave me the opportunity to delve into the fascinating world of drug repurposing. This blog post explores how identifying new uses for existing drugs can dramatically accelerate and reduce the cost of bringing vital treatments to patients, and how recent advances in machine learning can further streamline this process.

Building a Traffic Reminder Widget

TabPFN for chemical datasets

Deep Learning models have traditionally performed well on unstructured data such as text and images, but poorly on structured tabular data, and are usually outperformed by Gradient Boosted Decision Trees (GBDTs) on tabular chemical data. TabPFN (Tabular Prior-data Fitted Network) is a transformer-based foundation model for tabular data, pre-trained on millions of synthetic datasets to solve supervised learning tasks, with state-of-the-art performance on benchmarks. But does it work for cheminformatics?

Working with large virtual chemical libraries: Part 2 - Genetic algorithms

This is part 2 of a a planned three post series on working with large chemical libraries. The notebook used to create this post, and all the files can be found in this github repo.

Displaying distributions with raincloud plots

I’ve tried to visualise and compare distributions using violin plots for reports and presentations in the past, and the feedback I’ve got was generally… not great. When searching for better methods I came across this excellent blog post by Alex Belengeanu on raincloud plots and I’m now a big fan.

Working with large virtual chemical libraries: Part 1 - Active learning

This is part 1 of a planned three post series on working with large chemical libraries. The notebook used to create this post and all the files can be found in this github repo.

I want to become a data scientist, but I have no idea where to start...