|
Data Science Seminar Date: Friday, 12 April 2024 Time: 2:00 PM Venue: NKN Hall Statistics and Machine Learning in Protein Design Maunish Baravila and Anirudh Bhashyam PopVax, Hyderabad. 12-04-24 Abstract Protein design is an indispensable tool to design vaccines. It involves an intricate web of Physics, Mathematics and Biology. Physics governs how proteins fold, mathematics enables us to explore the space of proteins and biology constrains our designs to specific requirements. Here we present a brief introduction to the work PopVax conducts and some of the challenges we undertake on a daily basis analyzing and predicting data relating to proteins, lipids and nucleic acids that are crucial to the development of our vaccine platform. Biology relies on the extensive playground offered by statistics to inform and assess the quality and implications of experimental data. This may take the form of something simple such as testing data for normality to something as complex as assessing the significance of survival curves using the Mantel-Cox log rank test. At PopVax we produce immense amounts of data from different teams that involve multiple parameters and outcomes, from the characteristics of novel lipids to the neutralization of pseudoviruses from designed antigens. This not only requires careful analysis but intelligent procedures to be able to obtain meaningful insights that can relay key information to the different teams. And of course data analysis is incomplete without informative visualizations that are widely understood. About the speakers: Dr. Maunish Barvalia is VP of Platform Technologies, Head of Immunology & Nucleic Acid Delivery. He has a PhD in Microbiology and Immunology (UBC) focusing on single-cell technologies. He has worked with Dr. Pieter Cullis to utilise LNP delivery for immune cell targeting at UBC and NanoVation Therapeutics. Anirudh Bhashyam is Scientist Computational Biology. He has an M.Sc. Scientific Computing with a specialisation in computational geometry and statistics. He is working towards designing protein function guided tools for machine learning aided protein design.
|
