Dr Ke Yuan - AI for Cancer Research

Introduction

Vision 

Our lab is at the forefront of integrating artificial intelligence (AI) with cancer research. We aim to harness the power of AI to make biologically impactful predictions, drawing from the vast and rich datasets generated through cutting-edge cancer research. By focusing on building AI models across various biological data modalities, we seek to transform how cancer is diagnosed, understood, and treated, accelerating progress toward personalised medicine and therapeutic advancements. 

Key Research Areas 

1. AI for Histology and Spatial Deep Phenotyping 
   We are developing AI algorithms to analyse histological slides and integrate data across spatial biology and genomics. This will allow us to map the histomorphological landscape of both mouse models and human samples, improving phenotypic assessment and accelerating discoveries in tumour characterisation and treatment response. 

2. Large Language Models (LLMs) for Biological Sequences 
   We develop LLMs tailored to biological sequences such as proteins and RNA. These models allow us to predict the effects of genomic mutations, including rare events, providing new insights into protein-protein interactions and RNA functions in cancer biology. 

3. Comprehensive In Silico Tumor Models 
   We are building digital twin models of mouse systems, which simulate genotype-phenotype relationships to predict treatment responses and intervention outcomes. This initiative holds the potential to revolutionise preclinical cancer research by enabling real-time, cost-effective simulations of therapeutic strategies. 

Methodology 

We develop machine learning and deep learning models tailored to make novel inferences in cancer. We utilise state-of-the-art local and national computing infrastructures, large-scale datasets, and advanced spatial biology tools to ensure the robust training and validation of our models. 

Impact and Applications 

The application of our AI models has the potential to significantly advance the fields of cancer diagnosis, prognosis, and treatment. By integrating histopathology, spatial biology, and genomic data, we aim to produce AI-driven diagnostic tools that will enhance routine clinical workflows, personalise cancer treatments, identify new biomarkers for drug development, and deliver breakthroughs in understanding cancer's molecular underpinnings.  

Collaborations 

We collaborate very closely with world-class pathologists, computational biologists, and cancer researchers from the University of Glasgow, CRUK Scotland Institute, NHS Greater Glasgow and Clyde, and international institutions. These partnerships provide us with the expertise and resources to bridge the gap between computational innovation and clinical application, ensuring that our AI tools make a tangible impact on cancer research and patient care. 

Email