AI Seminar: Design and prediction of biological sequences with neural machines
At this seminar, Debora Marks will describe how she has developed new methods in unsupervised and semi-supervised machine learning to exploit an enormous natural diversity to design biological sequences. The seminar is relevant to researchers and students interested in machine learning and its application in biology.
Debora Marks, Associate Professor of Systems Biology at Harvard Medical School.
Design and prediction of biological sequences with neural machines
What can we do with millions or billions of genomes? There’s now an amazing opportunity to develop machine learning methods that can exploit this enormous natural diversity to design of biological sequences for therapeutics and biotechnology, predict the effects of human genetic variation, predict how viral proteins may evolve, dynamics, the effects of mutations and the molecular design of biological systems. I will describe how we have developed new methods in unsupervised and semi supervised learning - we can accurately predict the clinical effects of protein mutations and design smart nanobody libraries optimized for stability and diversity. I will introduce challenges for extending these methods to designing specific protein functions, biased library design and examples of probabilistic models to generate novel functional biotherapeutics.
This seminar is a part of the AI Seminar Series organised by SCIENCE AI Centre. The series highlights advances and challenges in research within Machine Learning, Data Science, and AI. Like the AI Centre itself, the seminar series has a broad scope, covering both new methodological contributions, ground-breaking applications, and impacts on society.