Aliee Group

Explain. Predict. Outsmart cancer.

Research Summary

We are developing artificial intelligence models to understand the molecular mechanisms driving health and disease. With these models, we will explore what makes cells unique, what makes individuals different, and how we can influence these differences to treat diseases.

Introduction

Our work aims to answer a lot of questions. What makes cells unique? What makes individuals different? And how can we influence these differences to treat diseases? Digging deeper we want too look at how cells make decisions and how these decisions drive changes during development or disease. More critically, how do these cellular behaviours vary across individuals? What principles govern how cells respond to their environment, including drugs and disease? And how do cells interact with each other, and how do these interactions shape their responses?

By addressing these questions, we want to understand how and what to change to guide a cell, and ultimately an organ or individual, toward a desired response. This knowledge will enable more personalised treatments and more inclusive, effective healthcare solutions.

To tackle these questions, we will develop AI-driven computational methods and modelling techniques tailored to the complexities of biology. Our central goal is to understand what drives cellular responses. For example, we aim to identify which drug, or mutation, induces a specific effect in a cell. Once our models demonstrate robust predictive performance, we will focus on their ability to answer counterfactual questions. For example, what is the likelihood of a cell returning to a healthy state if a disease-driving mutation, such as BRCA1 in breast cancer, is corrected in cells currently carrying it? This kind of counterfactual reasoning is critical for exploring disease reversal and guiding therapeutic interventions aimed at restoring health at the cellular level.

Reaching these goals requires more than just good data and well-tuned models; it also demands careful estimation of model uncertainty and generalizability. We aim to develop models that encode underlying biological mechanisms, providing both robust causal insights and the ability to generalize across diverse biological contexts.

Dr Hana Aliee profile picture

Dr Hana Aliee

Junior Group Leader

Research topics

AI-driven scientific discovery

Hypothesis formation for designing novel experiments

Generating hypotheses in biology is extremely difficult. Biological systems are immensely complex, involving countless interacting components, from genes and proteins to cells and tissues, operating in dynamic environments. This complexity, coupled with the high cost and time required for experiments, makes systematic exploration of biological questions extremely challenging.

To overcome these barriers, we aim to develop AI-driven models that can assist in formulating testable hypotheses and designing informative experiments. These models will learn from existing biological data and integrate new experimental results in a closed loop, continually refining their understanding of biological systems. By capturing patterns in data and simulating outcomes under various perturbations, our models will help prioritize experiments that are most likely to yield novel insights.
Our AI-based models will complement human expertise and support researchers in unlocking deeper understanding of living systems.

Modelling Cellular Intelligence

What makes a cell? From complexity to comprehension to hypothesis

From their DNA to the environment they inhabit, cells are complex living systems, shaped by a multitude of factors. Building reliable models of cells can help us understand how they work and, crucially, how they respond to different changes, such as genetic mutations, drugs or diseases.

Thanks to recent advances in AI, this once-distant goal is becoming increasingly possible. These models can generate hypotheses, guide experiments, and eventually enable us to design gene circuits, engineer cells and tissues and better understand and treat diseases.

In this project, we will develop AI-powered models that capture how and why cells differ across individuals and conditions. Our focus is on creating models that are not only predictive, but also generalizable—able to adapt to shifts caused by disease, environment, or other biological variation.

Principles of Gene Regulation Across Individuals and Contexts

Causality meets cell biology

In biology, causal questions often focus on understanding the link between a genetic variant and a disease phenotype, the effectiveness of a drug in a specific individual or the impact of a perturbation within a cell. These questions revolve around identifying the causal role of a specific trigger such as a mutation, drug, or genetic modification.

At the heart of these inquiries is the study of gene interactions, typically represented by a Gene Regulatory Network (GRN) and how these interactions shift in response to such triggers. Inferring GRNs is a complex challenge due to the stochastic nature of biological data and the inherent intricacies of gene regulation. These networks are highly context-dependent with gene interactions varying across tissue types, developmental stages and environmental conditions. Additionally, they often involve nonlinear and dynamic interactions that evolve over time and in response to external stimuli or disease states.

Our research aims to develop computational methods that address these challenges and enable more accurate, context-aware inference of GRNs. Beyond the technical challenges, we are driven by a deeper, philosophical question: what are the universal principles of gene regulation at the molecular level, analogous to the laws of gravity in physics, that hold across individuals? And, if such principles exist, how do they adapt or diverge in different environments or across genetically diverse populations? Exploring these questions could reveal foundational rules that govern life’s molecular logic, while also guiding personalized and equitable approaches to health and disease.