Meet the new researchers joining LSI

My current key research interest is centred on creating intelligent, adaptive systems capable of learning from experience, adapting to dynamic environments, and driving fundamental scientific discovery. My group is one of the most active places in the world for cutting-edge frontier AI. My current research is mainly in the following three pillars.

1. AI for scientific discovery: We are developing very large-scale foundation models to understand and programme biology. At the moment, my group is intensively working on molecular level, particularly the modeling of structure-mediated RNA functions (e.g., predicting RNA tertiary structures) and design and optimization of functional RNA sequences. Meanwhile, we are engineering AI agentic systems to work as virtual co-scientists side-by-side with human.
2. Evolutionary transfer optimisation: We are working on developing theoretical understanding and practical algorithms on evolutionary transfer optimisation, a new paradigm that enables optimisation to acquire experience from previous tasks, consolidate this knowledge, and reuse it for new, challenging problems. This makes optimisation significantly more efficient and effective, particularly in dynamic or resource-constrained environments.
3. Multi-objective optimisation for trustworthy AI: We are also explores how evolutionary transfer optimisation and multi-objective optimisation can enable trustworthy AI. For example, how can we efficiently optimize the performance of a deep neural network by tuning its hyperparameters? How can we generate effective adversarial examples to identify the vulnerabilities of deep learning systems? How can we optimize the token combinations to automatically generate injection cases to test web applications?

Currently, my group is working on three areas. One is very large-scale foundation models that learn regulatory code from massive genomics data, as well as its parameter efficient fine-tuning to adapt to a variety of downstream biological tasks with very limited experimental data. In addition, we are also working on explainable AI methods to analyze the decision patterns of these foundation models. We expect to generate symbolic rules that are not only grounded in the existing literature, but also falsifiable as new, unknown motifs ready for experimental validation. Last but not least, we are building AI agentic systems that work as co-scientists to expand the human scientists' capability and accelerate new scientific discovery.

I chose the LSI because it offers the unique interdisciplinary ecosystem required to bridge the gap between AI and complex living systems. I believe the complexity of living systems will create unprecedented new challenges to push the frontier AI, with new AI models, algorithms, and systems. On the flip side, I also look forward to strengthening collaborations with LSI colleagues to apply our technologies to address their frontier biological challenges, which go beyond their traditional pipelines.

I received my PhD from Stanford University in 1995 working on X chromosome inactivation with Huntington Willard. In 1995 I joined the lab of Adrian Bird at the University of Edinburgh and participated in the discovery and characterisation of a family of methyl-CpG binding proteins in mammals. In 2001 I started my own laboratory in the Centre for Genome Research at the University of Edinburgh where we investigated how control of chromatin and transcription is important in mammalian stem cells. I was then a group leader at the Cambridge Stem Cell Institute from 2008-2025, still focussed on control of transcription in stem cells, but also how mutations in genes encoding chromatin remodelling proteins leads to human genetic disease. This will continue to be our focus in Exeter from September 2025.

We will continue our collaborations with existing LSI groups to better understand how chromatin remodelling is used by human pluripotent stem cells to facilitate cell fate decisions. In addition, we are excited to expand our research in a number of areas in collaboration with other LSI and UoE groups. We will use advanced structural methods to understand how the activity of chromatin remodellers is influenced by protein interactions. We will continue to investigate how mutations in chromatin remodelling protein genes give rise to human genetic disease, and we will adopt and advance the latest methods used to dissect chromatin biology, gene expression and nuclear structure together with the wider University of Exeter chromatin community.

I chose to move my lab to the LSI because of the breadth of excellent research happening both in the LSI and across the University. I was struck by the very collaborative environment, and the ease with which one can pursue highly interdisciplinary collaborations and joint projects. All of this can be achieved in a beautiful setting with easy access to the natural beauty of Devon.