Hew Phipps

Banner

I'm Hew, a PhD student at the University of Oxford researching generative flow models for protein folding in the Oxford Protein Informatics Group (OPIG) supervised by Prof. Charlotte Deane.

I’m also a self-employed ML and software consultant with a portfolio of work for Isomerase where I developed their flagship directed evolution tool Evoselect.

Feel free to reach out to me if you’re interested in collaborating or my ML consulting.

✉ Email Me

Latest

Attending ICML 2026

Hew Phipps published on 2026-05-30

I will be attending ICML 2026 this year in Seoul Korea where I will be presenting my new paper ‘Spectral Volume Diffusion for Protein Dynamics’ at the GenBio Workshop.

Attending Roche Continents 2026

Hew Phipps published on 2026-05-30

I have been invited to attend Roche Continents 2026, an interdisciplinary merger of science and the arts in the beautiful historic town of Arles, France. I’m super thankfuol to Roche for this amazing opportunity and looking forward to a week of sun and stepping out of my comfort zone.

I Made an Obsidian Plugin That Turns Obsidian Into a Local LLM Agent IDE

Hew Phipps published on 2026-05-29

I made an Obsidian plugin for local Ollama agentic model chats within Obsidian through a neat sidebar. It’s optimised for local-first safe tool usage, agent configurations and cool things like automated internal note linking.

New Blog for OPIG: Spin Lattices and Proteins

Hew Phipps published on 2026-05-28

I wrote a new blog for my research group OPIG on how spin glass lattice models in Physics have enabled and continue to shape modern protein modelling

Modelling Protein Folding With Graph Networks Bridges Geometry and Kinetics

Hew Phipps published on 2026-01-20

Introduction

Protein folding is still a difficult problem. Although AlphaFold made great strides in advancing static structure prediction, the dynamic process that transports a protein from random coil to equilibrium structure is difficult to model. Go based Ising-like energy models such as the WSME model were among the first successful mathematical models of protein folding as a kinetic process. The approach is as follows: given a native-like folded conformation, we can assign residues with any possible alternative conformation as either native-like or not. One way to define this is to consider the residue’s formation of native-like contacts ($C_\alpha - C_\alpha$ distance $<8\dot{A}$). In folding this could be the difference between a random coil unfolded state where a given residue makes few native-like contacting pairs with spatially proximal residues, versus the folded state where that residue may exist in an alpha helix.

Flow Matching From the Mathematics

Hew Phipps published on 2025-10-01

Introduction

In the world of computational structural biology you might have heard of diffusion models as the current big thing in generative modelling. Diffusion models are great because primarily they look cool when you visualise the denoising process to generate a protein structure (checkout RFdiffusion Colab notebook), but also because they are state of the art at diverse and designable protein backbone structure generation.