Interactive code editors have been around for a while now, and tools like GitHub Copilot have woven their way into most development pipelines, and for good reason. They’re easy to use, exceptionally helpful (at certain tasks), and have undeniably made life as a developer smoother. Recently, I decided to switch away from relying on GitHub Copilot in favour of a local model for a few key reasons. While I don’t use it all the time, it has proven to be a useful option in many situations. In this blog post, I’ll go over why I made the switch, how I set it up, and share a bit about my experience so far.
Continue readingNarrowing the gap between machine learning scoring functions and free energy perturbation using augmented data
I’m delighted to report our collaboration (Ísak Valsson, Matthew Warren, Aniket Magarkar, Phil Biggin, & Charlotte Deane), on “Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data”, has been published in Nature’s Communications Chemistry (https://doi.org/10.1038/s42004-025-01428-y).
During his MSc dissertation project in the Department of Statistics, University of Oxford, OPIG member Ísak Valsson developed an attention-based GNN to predict protein-ligand binding affinity called “AEV-PLIG”. It featurizes a ligand’s atoms using Atomic Environment Vectors to describe the Protein-Ligand Interactions found in a 3D protein-ligand complex. AEV-PLIG is free and open source (BSD 3-Clause), available from GitHub at https://github.com/oxpig/AEV-PLIG, and forked at https://github.com/bigginlab/AEV-PLIG.
Estimating the Generalisability of Machine Learning Models in Drug Discovery
Machine learning (ML) has significantly advanced key computational tasks in drug discovery, including virtual screening, binding affinity prediction, protein-ligand structure prediction (co-folding), and docking. However, the extent to which these models generalise beyond their training data is often overestimated due to shortcomings in benchmarking datasets. Existing benchmarks frequently fail to account for similarities between the training and test sets, leading to inflated performance estimates. This issue is particularly pronounced in tasks where models tend to memorise training examples rather than learning generalisable biophysical principles. The figure below demonstrates two examples of model performance decreasing with increased dissimilarity between training and test data, for co-folding (left) and binding affinity prediction (right).
LLM Coding Tools – An Overview
We’ve come a long way since GitHub Copilot first showed us what AI-assisted coding could look like. These days, there’s a whole ecosystem of LLM coding tools out there, each with their own strengths and approaches. In this blog, I’ll give you a quick overview to help you figure out which one might work best for your workflow.
Level 1: Interactive Code Assistance
Continue reading
De novo protein padlocks
Binding a desired protein tightly is important for biotechnology. Recent advances in deep learning have allowed the de novo design of (mostly α-helical) binding protein, sidestepping the laborious process of raising antibodies or nanobodies or evolving affibodies, darpins or similar. These deep learning designed binders will bind with okay affinity, but what if the affinity required were much stronger?
<Enter autocatalytic isopeptide bonds>
Molecule Networks: data visualization using PyVis
Over the past few years I have explored different data visualization strategies with the goal of rapidly communicating information to medicinal chemists. I have recently fallen in love with “molecule networks” as an intuitive and interactive data visualization strategy. This blog gives a brief tutorial on how to start generating your own molecule networks.
Baby’s First NeurIPS: A Survival Guide for Conference Newbies
There’s something very surreal about stepping into your first major machine learning conference: suddenly, all those GitHub usernames, paper authors, and protagonists of heated twitter spats become real people, the hallways are buzzing with discussions of papers you’ve been meaning to read, and somehow there are 17,000 other people trying to navigate it all alongside you. That was my experience at NeurIPS this year, and despite feeling like a microplankton in an ocean of ML research, I had a grand time. While some of this success was pure luck, much of it came down to excellent advice from the group’s ML conference veterans and lessons learned through trial and error. So, before the details fade into a blur of posters and coffee breaks, here’s my guide to making the most of your first major ML conference.
Continue readingDiagnostics on the Cutting Edge, Software in the Stone Age: A Microbiology Story
The need to treat and control infectious diseases has challenged humanity for millennia, driving a series of remarkable advancements in diagnostic tools and techniques. One of the earliest known legal texts, the Code of Hammurabi, references the visual and tactile diagnosis of leprosy. For centuries, the distinct smell of infected wounds was used to identify gangrene, and in Ancient Greece and Rome, the balance of the four humors (blood, phlegm, black bile, and yellow bile) was a central theory in diagnosing infections.
The invention of the compound microscope in 1590 by Hans and Zacharias Janssen, and its refinements by Robert Hooke and Antonie van Leeuwenhoek, marked a turning point as it enabled the direct observation of microorganisms, thereby linking diseases to their microbial origins. Louis Pasteur’s introduction of liquid media aided Joseph Lister in identifying microbes as the source of surgical infections, whilst Robert Koch’s experiments with Bacillus anthracis firmly established the connection between specific microbes and diseases.
Continue readingMaking pretty, interactive graphs the simple way – Use Plotly.
Using an ESP8266 and some DS18B20 one-wire temperature sensors, I have been automatically recording temperature data from various parts of my pond, to see how it fluctuated with air temperature, depth and filter configuration.
Despite the help I was receiving from the feline fish monitor, I was getting a bit irked at the quality of the graphs I was getting using matplotlib.
Matplotlib has been around since 2003, more than 20 years now. It’s arguably the defacto method of producing graphs in python and it’s not going away. However, it’s also a pain to use and by default produces some quite ugly plots unless you put in the mileage. In fact, when attempting to quickly explore data, Michael L. Waskom’s frustrations with matplotlib were directly related to the production of the seaborn library. “By producing complete graphics from a single function call with minimal
arguments, seaborn facilitates rapid prototyping and exploratory data analysis.”
Seaborn makes use of matplotlib and integrates tightly with pandas provide a neat wrapper for matplotlib functions, allowing you to avoid a lot of the data herding needed to view a graph.
You may think “OK, so seaborn finally tames matplotlib, why should I use anything else?” In short, interactivity. Seaborn and Matplotlib may produce graphs, but a graph alone doesn’t really let you explore the data. If you look at a graph you’re limited to the scale the author thought made sense, you can’t zoom in or out and if one line is behind another, you’re kind of stuck.
Where plotly really shines is with just two lines you can generate your figure and then either save it as the image below, or as an interactive HTML graph such as this.
A tougher molecular data split – spectral split
Scaffold splits have been widely used in molecular machine learning which involves identifying chemical scaffolds in the data set and ensuring scaffolds present in the train and test sets do not overlap. However, two very similar molecules can have differing scaffolds. In an example provided by Pat Walters in his article on splitting chemical data last month, he provides an example where two molecules just differ by a single atom and thus have a very high Tanimoto similarity score of 0.66. However, they have different scaffolds (figure below).
In this case, if one of the molecules were in the train set and the other in the test set, predicting the test molecule would be quite trivial as there is data leakage. Therefore, we need a better splitting method such that there is minimal overlap between the train and test set. In this blogpost, I will be discussing spectral split, a splitting method introduced by our fellow OPIG member, Klarner et. al (2023).
Spectral split
Spectral split or clustering is based on the spectral graph partitioning algorithm. The basic idea of spectral clustering is as follows: The dataset is projected on a R^n matrix. An affinity matrix using a kernel that could be domain-specific is defined. Following that, the graph Laplacian is computed from the affinity matrix, followed by its eigendecomposition. Then, k eigenvectors corresponding to the k lowest/highest eigenvalues are selected. Finally, the clusters are formed using k-means.
In the context of molecular data splitting, one could use the Tanimoto similarity metric to construct a similarity matrix between all the molecules in the dataset. Then, a spectral clustering method could be used to partition the similarity matrix such that the similarity within the cluster is maximized whereas the similarity between the clusters is minimized. Spectral split showed the least overlap between train (blue) and test (red) set molecules compared to scaffold splits (figure from Klarner at. al. (2024) below)
In addition to spectral splits, one could attempt other tougher splits one could attempt such as UMAP splits suggested by Guo et. al. (2024). For a detailed comparison between UMAP splits and other commonly used splits please refer to Pat Walters’ article on splitting chemical data.