Abstract

Drug discovery is akin to finding a needle in a haystack: There are, theoretically, around 1 million million stable organic molecules. However, out of those theoretically stable molecules, we only know how to make around 0.1% them, and it is experimentally infeasible to test every molecule that we know how to make. In my talk, I will discuss machine learning algorithms that can predict which molecules are likely to be functional against medically physiological processes by learning from the published chemiDrug discovery is akin to finding a needle in a haystack: There are, theoretically, around 1 million million stable organic molecules. However, out of those theoretically stable molecules, we only know how to make around 0.1% them, and it is experimentally infeasible to test every molecule that we know how to make. In my talk, I will discuss machine learning algorithms that can predict which molecules are likely to be functional against medically physiological processes by learning from the published chemical literature, as well as predicting ways to make those active molecules from readily available starting materials. I will show that the design of machine learning algorithms invites us to think about chemistry and molecules in a fundamentally different way. cal literature, as well as predicting ways to make those active molecules from readily available starting materials. I will show that the design of machine learning algorithms invites us to think about chemistry and molecules in a fundamentally different way.

The talk is open to the public!

Start: 18:00 1hr talk followed by a Q&A, then cheese & wine & a chat!

Where: Pfizer lecture theatre, Chemistry department

Members: Free | Non-members: £3

https://www.facebook.com/events/1896620497094852/