Mapping protein motion with generative field models
Dr. Lena Okafor joined Corvela Labs in January 2026 as a Research Scientist in the Molecular Dynamics and Generative Modelling (MDGM) Team.
Before joining Corvela Labs I had spent years studying how proteins flex and fold, and I was fascinated by how much of a molecule's behaviour lives in its motion rather than in any single static structure. I also knew how slow and costly it can be to simulate those dynamics, but I felt the tools were finally reaching a turning point.
When I started at Corvela, I was struck by how far the generative field models had come and by the freedom they gave us to test ideas quickly. For me the impact was clear across a range of problems — for example, they let us sample plausible conformational states for a target in minutes rather than weeks.
Models that support a multi-disciplinary team
Bringing new medicines to patients is a long and uncertain process, and success is never guaranteed. Corvela Labs takes a distinctive approach, working from first principles to answer the fundamental questions that reshape how we design drugs. Our goal is to make the process more efficient, more reliable and safer, so we can reach difficult disease areas with fewer side effects.
To do this we've built a multi-disciplinary drug-design team that brings together medicinal chemists, computational chemists and structural biologists working alongside machine learning researchers, engineers and software developers. Our culture places real emphasis on sharing knowledge across disciplines and cross-pollinating ideas.
We also believe that to answer the hardest questions about disease and human health, we need a family of predictive and generative models that support scientists across many kinds of expertise.
Working from different starting points
Developing and iterating on Corvela's field models has been central to our progress. Predicting how a ligand and a protein move together is a critical part of discovery, because it reveals the intricate networks of interactions that govern how molecules inhibit or modulate a protein's function.
Because scientists arrive from different backgrounds, the same model is read in many ways. A chemist and a biologist will interrogate a predicted trajectory with different instincts, and that diversity of thinking is exactly what makes the results more robust.
Looking ahead
Beyond the small-molecule paradigm, these models have the potential to support new therapeutic approaches based on modulating protein–protein and protein–nucleic acid interactions. With access to large-scale compute we can deploy them at scale and interrogate many hypotheses at once, and I believe this is where the next decade of design will be won.