Artificial Intelligence for drug discovery GTC Europe 2017 10 - PowerPoint PPT Presentation

Artificial Intelligence for drug discovery GTC Europe 2017 10 October 2017 Dean Plumbley Senior Machine Learning Scientist

Table of contents 1. About us 2. Hypothosis Generation 3. Current Success! 4. Drug Design using AI 5. AutoChem – Chemical Property Prediction 6. EvoChem – Drug generation 7. Looking Ahead

We are Since our foundation in 2013, our mission has been to bring together the best of technology and scientific research to enable us to create better medicines. BenevolentAI harnesses artificial intelligence to enhance and accelerate scientific discovery by making sense of highly fragmented information to create new insights and usable knowledge that benefit society.

BenevolentAI Overview Founded in 2013 $BN company 80+ team of world class scientists and technologists (50+ doctorates/advanced degrees) Technology enabling previously impossible tasks in bioscience Rich patent portfolio of over 400 patents 10/10/2017 4

Drug Discovery is broken! 10/10/2017 5

BenevolentAI is challenging traditional approaches 10/10/2017 6

Proprietary Knowledge and Inference Models 10/10/2017 7

BenevolentAI finds potential treatments for ALS 10/10/2017 8

AI driven compound design

Overview of AI driven compound design 10/10/2017 10

Drug design using AI Property Prediction Molecule Generation MoleculeNet: A Benchmark for Molecular Machine Learning Wu et al, arXiv 2017, https://arxiv.org/abs/1703.00564 10/10/2017 11

AutoChem Drug property prediction

QSAR modelling Drug structures as SMILES Representation as fingerprints / physchem descriptors 10/10/2017 13

QSAR modelling Low Data Drug Discovery with One-shot Learning Altae-Tran et al, 2016, https://arxiv.org/abs/1611.03199 10/10/2017 14

AutoChem – The system 10/10/2017 15

EvoChem Generative Chemistry

Bespoke molecule generation • Multiparameter optimisation of drugs is hard! • Can AI help? Diversity-Oriented Synthesis: Developing New Chemical Tools to Probe and Modulate Biological Systems Galloway et al, 2014, http://www- spring.ch.cam.ac.uk/publications/pdf/2014_DOS_379.pdf 10/10/2017 17

Generative models for drug design Automatic chemical design using a data-driven continuous representation of molecules Gomez-Bombarelli et al, 2016, https://arxiv.org/abs/1610.02415 10/10/2017 18

Thanks for listening! Any Questions? We’re hiring! Get in touch Dean.Plumbley@benevolent.ai