Skip to main content

MATCHMAKERS

Led by Dr Michael Birnbaum (Massachusetts Institute of Technology, USA).

The challenge

MATCHMAKERS are tackling the T-cell receptors challenge. The team aim to crack the code of how T cells recognise cancer cells by harnessing recent advances in structural biology and genetic screening as well as molecular modelling, computational biology and artificial intelligence (AI). By generating and integrating vast datasets, they plan to develop powerful new tools that can predict T cell antigen specificity – paving the way toward the development of more effective and personalised immunotherapies.

Remote video URL

Why T-cell receptors?

Hear Dr Michael Birnbaum, team lead of MATCHMAKERS, discuss their approach to the T-cell receptors challenge - and why addressing it is so important.

At a glance

Funded in 2024
Institutions 10
Locations Brazil, Germany, the Netherlands, Norway, UK, US
Specialisms cancer immunology, computer science, high-throughput method development, structural biology

Team and leadership

MATCHMAKERS is led by:

Dr Michael Birnbaum

This international team of researchers brings together world-leaders in cancer immunology, computer science, high-throughput method development and structural biology, from 10 institutions across five countries.

Tackling the T-cell receptors challenge

Cutting-edge immunotherapies that harness the power of T cells are transforming cancer treatment, but their effectiveness varies across cancer types and patients. Understanding how T cells recognise and destroy cancer cells is crucial for extending these benefits to more patients. T cells recognise cancer not through direct antigen binding, but by detecting peptides displayed on the cell surface by major histocompatibility complex (MHC) molecules. Each T cell has a unique T-cell receptor (TCR), capable of recognising many peptide-MHC (pMHC) combinations through extensive cross-reactivity. Decoding how TCRs recognise tumour antigens presented by MHCs is critical to advancing personalised immunotherapy. Yet, the enormous diversity and cross-reactivity of TCRs has made it incredibly challenging to determine which tumour antigens T cells recognise, how these interactions occur, and why responses to immunotherapy differ between patients.

To tackle this challenge, MATCHMAKERS will generate extremely large, integrated datasets of TCRs and their target pMHCs and use these to build computer models to accurately predict antigen recognition by different TCRs. The team’s overarching goal is to allow accurate prediction of T-cell recognition in a clinical setting using routine tumour biopsies or blood samples . Additionally, they will leverage their knowledge of TCR-pMHC interactions to design antigen-specific TCRs for personalised immunotherapies.

A major challenge in using computational methods to predict TCR-pMHC interactions has been the vast amount of data required to train algorithms effectively. However, recent advances in high-throughput approaches have now made it possible to map interactions between TCRs and antigens at an unprecedented scale. The team has developed a scalable TCR rapid assembly platform, T-RAP, that allowed the generation of large and uniform TCR libraries. T-RAP has enabled evaluation of TCR signalling or MHC multimer binding of thousands of previously reported TCR-pMHC pairs, providing a path towards faster selection of tumour-reactive receptors for personalised TCR therapies. The team has also developed TCRAFT, a rapid, low-cost method for assembling large TCR libraries and paired it with functional antigen screening to map reactivity and screen over 30,000 TCRs from pancreatic cancer donors. This addresses a major scalability bottleneck moving from receptor sequences to experimentally confirmed antigen recognition and could accelerate therapeutic TCR discovery, while also revealing shared features of antigen specific T cells in autoimmunity and cancer. The sheer volume of complex data, and ability to validate their models, will allow the team to develop computational tools to accurately predict, for the first time, unseen patient-specific tumour antigens recognised by different TCRs, and to design new TCRs for personalised immunotherapies.

The team has also demonstrated how AI-guided protein design and SMART MHCs can enable high-throughout, crystallisable, and generalisable antigen presentational tools. AI-designed, high specific binders to peptide-MHC-1 complexes offer a scalable route to more precise immunotherapies and diagnostics, potentially expanding access to intracellular cancer targets while sparing healthy tissue. This marks a shift from discovering natural TCRs to engineering antigen recognition, with rigorous cross-reactivity testing remaining essential for clinical translation.

The team has developed Boltz-2 and BoltzGen, highlighting a major step forward in data-driven biomolecular design by unifying accurate binding affinity prediction with generative binder creation at scale. By combining efficient modelling with generative approaches that generalise across targets, these open access tools significantly accelerate the discovery and optimisation of molecular binders, with broad implications for drug discovery, protein engineering, and therapeutic design.

Harnessing AI through Project AURORA

In 2026, MATCHMAKERS was selected to take part in Project AURORA, a pioneering Cancer Grand Challenges initiative exploring how AI agents could help research teams generate new hypotheses, design experiments and accelerate progress against cancer.

Led by Welles Robinson and Sergio Quezada at University College London, the MATCHMAKERS AURORA pioneer team is developing two complementary AI co-scientists. One will investigate which tumour antigens can trigger T-cell responses; the other will examine the features of T cells that are capable of recognising tumours. Together, they will help prioritise antigens and TCRs for experimental testing, supporting the team’s efforts to decipher how T cells recognise cancer.

The first will investigate what makes a tumour antigen capable of triggering a T-cell response. The second will examine the features of tumour-reactive T cells, including how their activity relates to the antigens a tumour displays. The resulting data could help MATCHMAKERS decipher the rules of tumour recognition and support the future development of personalised immunotherapies.

MATCHMAKERS' core programme is funded by Cancer Research UK, the National Cancer Institute and The Mark Foundation for Cancer Research through Cancer Grand Challenges. The AURORA project is funded by Cancer Research UK.

Welles and Marinka
Pictured left to right: Welles Robinson (Team MATCHMAKERS), Marinka Zitnik (Project Lead, Project AURORA)

Looking ahead

MATCHMAKERS aims to revolutionise the understanding of how an individual’s T cells recognise and target their tumour cells. If successful, their discoveries could transition the prediction of TCR antigen recognition from a highly specialised process limited to a few laboratories worldwide to a routine clinical application available for all. This knowledge will transform the ability to develop personalised immunotherapies tailored to the unique characteristics of an individual’s cancer and T cell repertoire.

Gaining a deeper understanding of TCR-antigen interactions will also have far-reaching implications beyond cancer– wherever antigen recognition plays a critical role, including infectious diseases, autoimmunity and allergies.

Michael Birnbaum
Our team has the combination of immunologists, computer scientists and engineers needed to create the technologies to make it possible to predict what a T cell recognises, which will enable the development patient-specific TCR immunotherapies.
Michael Birnbaum (Team Lead, MATCHMAKERS)
MIT

Plain language summary

In the past decade, there has been a revolution in the treatment of some cancers using therapies that harness a person’s immune system. These therapies, called immunotherapies, use the power of T cells – a type of white blood cell that are key players in the immune response to infection and cancer. T cells have receptors on their surfaces that recognise ‘antigen’ molecules that come from pathogens or cancer cells. After a T cell attaches to a cancer antigen, the cancer cell is ‘marked’ for destruction by the immune system. 

Each of the millions of T cells in each person has a unique T-cell receptor that recognises different antigens. This diversity has limited our ability to know what exactly T cells recognise in the people who respond to treatment. If we can predict what cancer antigens a T cell ‘sees’ based on its receptor, we could find ways to create immunotherapies specific to each individual’s cancer and increase the overall effectiveness of these therapies across cancer types.  

To bring us closer to this goal, MATCHMAKERS will collect data on T-cell receptors and the different antigens they interact with. Using this data and recent advances in artificial intelligence, the team will build computer models to predict the antigens that are recognised by different T-cell receptors. These predictions will be used to design T-cell receptors that are specific to an individual’s cancer. 

If successful, MATCHMAKERS could transform the understanding of how an individual’s T cells recognise cancer cells from something that now can only happen in a few laboratories around the world, for a few people at a time, into a routine process. This will allow for better matching of patients with therapies that could lead to durable responses including remission and cure. 

Members of team MATCHMAKERS