EP 161: How large language models can help build immunotherapies with Michelle Teng of Etcembly Ltd.
November 14, 2024
In this week’s episode, Patrick is joined by Michelle Teng, CEO and Co-Founder of Etcembly Ltd and Founding Executive Director and Chief Scientific Officer of SynaptixBio. Michelle explains how her team is training large language models (LLMs) to analyze immune system data and how the company’s Long Term Survivor Study is helping identify T-cell profiles associated with sustained cancer remission. She also speaks to her own experience of ultra-rare genetic disease.
0:00 Intro to The Genetics Podcast
01:00 Welcome to Michelle
01:35 Explaining immunotherapy and its evolution over the past decade
04:10 Current insights on immunotherapy responders and the underlying factors driving varied individual responses
05:50 The latest generation of T-cell receptor therapies
08:53 The origin of the Long Term Survivor Study, its purpose and how it informs discovery of new T-cell receptor therapies
12:32 How Etcembly is characterising T-cells and antibodies in survivor profiles
15:00 Using machine learning to understand the immune system
18:44 The complexity of Human Leukocyte Antigen (HLA) and how it relates to differences in T-cell receptor biology
22:35 T-cell repertoires in Long Term Survivor Study participants
26:06 Training LLMs in immune system biology, data and more
27:54 Michelle’s work at Immunocore and how she’s applied her knowledge to grow Etcembly
33:02 Setting up a new company at the crossroads of the Covid-19 pandemic and the inception of LLMs and AlphaFold
35:54 Current bottlenecks in pre-clinical immunotherapy development
37:23 Michelle’s eldest daughter’s experience with an ultra-rare genetic disease and the founding of SynaptixBio
43:34 Utilising biobanks and registries to better understand ultra-rare disease presentation
47:05 The power and importance of patient parent groups for developing rare disease treatments
48:48 Closing remarks
Find out more