AI3SD Video: What a medicinal chemist needs to know about explainable artificial intelligence
AI3SD Video: What a medicinal chemist needs to know about explainable artificial intelligence
The latest developments in artificial intelligence (AI) have arrived into an existing state of creative tension between computational and medicinal chemists. At their most productive, medicinal and computational chemists have made significant progress in delivering new therapeutic agents into the clinic. However, the relationship between these communities has the prospect of being weakened by application of over-simplistic AI methods which, if they fail to deliver, will reinforce unproductive prejudices. AI systems are action orientated; they suggest, and even automate, possible steps to take next in drug hunting projects. They do this by generating options, preforming predictions, then ranking the options. A key piece of critical learning is that any AI system for chemists should be open and ‘explainable’. The requirement for the chemist to understand how models have been built and ‘drill back’ to original data is key to explain how the computer has arrived at the prediction. For example, an AI system that aids the medicinal chemist in evaluating and designing new biologically active compounds should focus on communicating, in the preferred mode of the medicinal chemist, via critical substructures, and map to the original compounds and test data on which the model was built. Without this, any model becomes labelled as ‘black-box’ and confidence is reduced in the suggestions made. Therefore, to develop them further and to understand the quality of any prediction, transparency and auditability should be designed into a system from inception (see example below where the predictions are made for a compound and the contributions from the model highlighted). The talk will show practical examples from drug discovery projects and recent compounds from the global Covid Moonshot drug discovery program.
AI, AI3SD Event, Artificial Intelligence, Chemistry, Machine Learning, Medicinal Chemistry
Dossetter, Alexander G.
d0ea65a2-fdd2-47f0-86d4-cce9a8b6148a
Frey, Jeremy G.
ba60c559-c4af-44f1-87e6-ce69819bf23f
Kanza, Samantha
b73bcf34-3ff8-4691-bd09-aa657dcff420
Niranjan, Mahesan
5cbaeea8-7288-4b55-a89c-c43d212ddd4f
24 November 2021
Dossetter, Alexander G.
d0ea65a2-fdd2-47f0-86d4-cce9a8b6148a
Frey, Jeremy G.
ba60c559-c4af-44f1-87e6-ce69819bf23f
Kanza, Samantha
b73bcf34-3ff8-4691-bd09-aa657dcff420
Niranjan, Mahesan
5cbaeea8-7288-4b55-a89c-c43d212ddd4f
Dossetter, Alexander G.
(2021)
AI3SD Video: What a medicinal chemist needs to know about explainable artificial intelligence.
Frey, Jeremy G., Kanza, Samantha and Niranjan, Mahesan
(eds.)
AI3SD Autumn Seminar Series 2021.
13 Oct - 15 Dec 2021.
(doi:10.5258/SOTON/AI3SD0168).
Record type:
Conference or Workshop Item
(Other)
Abstract
The latest developments in artificial intelligence (AI) have arrived into an existing state of creative tension between computational and medicinal chemists. At their most productive, medicinal and computational chemists have made significant progress in delivering new therapeutic agents into the clinic. However, the relationship between these communities has the prospect of being weakened by application of over-simplistic AI methods which, if they fail to deliver, will reinforce unproductive prejudices. AI systems are action orientated; they suggest, and even automate, possible steps to take next in drug hunting projects. They do this by generating options, preforming predictions, then ranking the options. A key piece of critical learning is that any AI system for chemists should be open and ‘explainable’. The requirement for the chemist to understand how models have been built and ‘drill back’ to original data is key to explain how the computer has arrived at the prediction. For example, an AI system that aids the medicinal chemist in evaluating and designing new biologically active compounds should focus on communicating, in the preferred mode of the medicinal chemist, via critical substructures, and map to the original compounds and test data on which the model was built. Without this, any model becomes labelled as ‘black-box’ and confidence is reduced in the suggestions made. Therefore, to develop them further and to understand the quality of any prediction, transparency and auditability should be designed into a system from inception (see example below where the predictions are made for a compound and the contributions from the model highlighted). The talk will show practical examples from drug discovery projects and recent compounds from the global Covid Moonshot drug discovery program.
Video
AI3SDAutumnSeminar-171121-AlDossetter
- Version of Record
Text
17112021-AI3SDQA-AGD
More information
Published date: 24 November 2021
Additional Information:
In 2012 Dr Al Dossetter co-founded MedChemica Limited centred around the technology of Matched Molecular Pair Analysis (MMPA) as a method of accelerating medicinal chemistry. MedChemica now licenses a suite of Artificial Intelligence databases, and tools on-line, for organisations to extract and share knowledge from their own data. The software and methodologies have been used by chemists in many pharmaceutical companies, universities and bio-techs to accelerate drug discovery programmes. Extending the methodology enabled MedChemica Limited to share medicinal chemistry knowledge between the research branches of AstraZeneca, Hoffman La Roche and Genentech. In addition MedChemica offers consultancy services on drug discovery project and Al has helped multiple project achieve their goals. Previous to MedChemica Al gained his PhD from Nottingham University and after post-doctoral research at Harvard University joined AstraZeneca (AZ). He spent 13 years in medicinal chemistry spread across oncology (hormonal and kinase inhibitors), inflammation (OA and RA, enzyme inhibitors and GPCR targets) and diabetes (obesity, GPCR and enzyme inhibitors), delivering multiple projects and candidate drugs.
Venue - Dates:
AI3SD Autumn Seminar Series 2021, 2021-10-13 - 2021-12-15
Keywords:
AI, AI3SD Event, Artificial Intelligence, Chemistry, Machine Learning, Medicinal Chemistry
Identifiers
Local EPrints ID: 453338
URI: http://eprints.soton.ac.uk/id/eprint/453338
PURE UUID: 3620f2de-cf06-4dc2-a294-f75a88d2cfd2
Catalogue record
Date deposited: 13 Jan 2022 17:40
Last modified: 17 Mar 2024 03:51
Export record
Altmetrics
Contributors
Author:
Alexander G. Dossetter
Editor:
Mahesan Niranjan
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics