In this post I’ll walk through how to set up the CCDC Python API and use the CSD Geometry Analyser to evaluate the geometric quality of molecules from three representative structure-based de novo design models. I’ve put together a small GitHub repo with the full analysis code where we look at bond lengths, angles, torsions, and ring conformations across the three methods, and compare these against their PoseBusters validity scores to see what each metric is really capturing.
Continue readingCategory Archives: Protein Structure
Spin Lattices and Proteins – How state-based discretisations have enabled modern protein modelling
I got into protein modelling not long before AlphaFold2 first released. At that time some of the prevailing methods for protein structure prediction came from highly interpretable energy functionals that arose from a particularly beautiful intersection of statistical mechanics and biology. These “Potts” models are going to be the centre of a larger discussion in this blog on state-based discretisations of proteins, how they’ve shaped modern deep learning methods and whether there is still more to learn from them.
In the age of black box deep learning, does the Potts model still have a place?
The Potts/Ising Model
The Ising model is a well established popular theoretical physics model of ferromagnetism. Simply put, given a lattice of atoms each capable of adopting 1 of 2 spins (up and down) ferromagnetism arises when their spins align and their associated magnetic moments point in the same direction. The Ising model tries to parameterise the local and non-local relationships between atoms and their spin states such that we can learn the Hamiltonian of the system and its different configurations under the magnetic field. The Hamiltonian takes the following form for a system of N atoms
$$
E = -\sum_{i}^Nh_ix_i – \sum_{i<j}^N J_{ij}x_i x_j,
$$
where J is the “coupling energy” between any two atoms x_i and x_j, and h represents the magnetic field, or more appropriately for our purposes it can be framed as a single-site field dictating how an individual atom independently acts within the model. You might recognise the form this binary spin model takes as it arises naturally across the sciences including in Hopfield networks and graphical models.
Everything is an Ising-like model if you’re brave enough
Continue readingA Golden Age of Nanomedicine
As someone who spent their entire academic career, from B.Sc. to M.Sc. to Ph.D., within a Kavli Institute for Nanoscience Discovery (first in Delft and now in Oxford), I’ve had the privilege of seeing firsthand just how beautifully intricate the nanoscale world can be. Now, as my research focuses on lipid nanoparticles for genetic therapeutics and vaccines, I would like to use this platform to advocate for what I believe is one of the most transformative frontiers in modern medicine: the rational design of nanomaterials for therapeutic delivery.
Continue readingAnalyzing AlphaFold 3’s Diffusion Trajectory
A useful way to understand AlphaFold 3’s sampling behavior is to look not only at the final predicted structure, but at what happens along the reverse diffusion trajectory itself. If we track quantities such as the physical energy of samples, noise scale, and update magnitude over time, a very clear pattern emerges: structures remain physically imperfect for most of sampling, and only take proper global shape in the final low-noise steps.
This behavior is a result of the diffusion procedure implemented in Algorithm 18, Sample Diffusion, which follows an EDM-style sampler with churn. Rather than simply marching monotonically from noise to structure, the sampler repeatedly perturbs the current coordinates, denoises them, and then takes a Euler-like update step. Because of the churn mechanism, AlphaFold 3 deliberately injects additional noise during part of the trajectory, which encourages exploration but also delays local geometric convergence. This mechanism is shown in step 4 -7 of the Sample Diffusion Algorithm from Alphafold3 Supplementary Information.
Misconduct, Bias or Benign? A Case of Missing Ångströms
An Ångström
An Ångström (Å) is a unit of length equal to 10−10 metres; one ten-billionth of a metre. It sits at a comfortable scale for the atomic world, with the diameter of a hydrogen atom, the length of a chemical bond, all measured in Ångström.
It is not an International System of Units (Système International d’Unités) “SI” unit. In fact, it has been formally deprecated in favour of the nanometre (1 Å = 0.1 nm), and standards bodies such as NIST and the BIPM discourage its use. Yet, in structural biology and chemistry, crystallography, and materials science, the Ångström persists. I would say, partly out of stubbornness, but mostly out of convenience. Saying a protein structure was solved at 2.1 Å feels natural in a way that 0.21 nm does not.
So we keep using it. And because we keep using it, we inherit its quirks and history.
Continue readingCan we make Boltz predict allosteric binding?
(While this post is meant to shed light on the problem of making AI structure prediction models like Boltz become better for allosteric binding, it is also an open call for collaborating on this problem.)
I recently took part in a Boltz hackathon organised by the MIT Jameel Clinic. I worked on improving Boltz 2 predictions for allosteric binders. The validation dataset provided was from a recent paper, Co-folding, the future of docking – prediction of allosteric and orthosteric ligands, which benchmarks some of the recent state-of-the-art AI structure prediction models on a curated set of allosteric and orthosteric binders. Generally, all AI structure prediction models are trained mostly on orthosteric binding cases, which means that their performance on allosteric binding is significantly worse.
Continue readingControlling the Diffusion Denoising Process: A Molecular Show
This blog post is supporting my poster at Young Modellers Forum and makes things way easier to see and understand. Underneath each GIF, is the explanation of what you should look for as things denoise throughout the diffusion trajectory. Click the GIFs for higher quality viewing!
Continue readingIs attention all you need for protein folding?
Researchers from Apple have released SimpleFold, a protein structure prediction model which uses exclusively standard Transformer layers. The results seem to show that SimpleFold is a little less accurate than methods such as AlphaFold2, but much faster and easier to integrate into standard LLM-like workflows. SimpleFold also shows very good scaling performance, in line with other Transformer models like ESM2. So what is powering this seemingly simple development?
Continue readingExploring the Protein Data Bank programmatically
The Worldwide Protein Data Bank (wwPDB or just the PDB to its friends) is a key resource for structural biology, providing a single central repository of protein and nucleic acid structure data. Most researchers interact with the PDB either by downloading and parsing individual entries as mmCIF files (or as legacy PDB files), or by downloading aggregated data, such as the RCSB‘s collection in a single FASTA file of all polymer entity sequences. All too often, researchers end up laboriously writing their own file parsers to digest these files. In recent years though, more sophisticated tools have been made available that make it much easier to access only the data that you need.
Continue readingAccelerating AlphaFold 3 for high-throughput structure prediction
Introduction
Recently, I have been conducting a project in which I need to predict the structures of a dataset comprising a few thousand protein sequences using AlphaFold 3. Taking a naive approach, it was taking an hour or two per entry to get a predicted structure. With a few thousand structures, it seemed that it would take months to be able to run…
In this blog post, I will go through some tips I found to help accelerate the structure predictions and make all of the predictions I needed in under a week. In general, following the tips in the AlphaFold 3 performance documentation is a useful starting place. Most of the tips I provide are related to accelerating the MSA generation portion of the predictions because this was the biggest bottleneck in my case.
Continue reading