by Bence Bruncsics (9:30)
Multivalent proteins provide a great tool in nature or biotechnology to increase binding strength and specificity by multiplying binding domains. Multiplying existing binding domains is a cost-effective way of improving binding properties relative to evolving or developing novel protein structures. We will discuss the development of the multivalent framework and present the binding kinetics in a system of differential equations, focusing on the application of math to biological systems. We can rewrite the ordinary differential equations (ODEs) in a matrix form and translate the results into a biologically relevant graph while focusing on showing the practical applications of mathematical tools.
Next, we will explore the concept of effective concentrations, which is an intrinsic property of multivalent complexes. We will use relatively simple mathematical concepts and we will discuss the process of finding out how to calculate the binding probability using integrals and probability density functions in a relatively intuitive manner.
Last, we will see some applications for multivalent models and discuss where similar tools or concepts can be used.
Molecule property prediction, federated learning and uncertainty quantification
by Lewis Mervin (14:00)
This talk explores the use of molecule property prediction approaches within the Design-Make-Test-Analyse (DMTA) cycle at AstraZeneca. We also outline how federated learning and uncertainty quantification are aiding drug design in-house.