How does generative chemistry work, and how can it help me?
The role of generative chemistry in drug discovery A key difficulty in finding new drugs is the sheer size of…
This is a common topic of conversation in the industry, and to get a true picture it’s important to separate hype from reality.
Is AI-guided drug discovery faster and cheaper?
The evidence for this is, by definition, anecdotal. No one runs the same project twice, with and without support from AI. And, it is inconceivable to do this a statistically significant number of times.
However, a growing number of anecdotal examples demonstrate substantial savings of time and effort in discovering candidate drugs. These add up to convincing evidence of AI’s potential. Some of these are retrospective analyses, exploring the same chemical space of a project previously prosecuted with traditional methods and showing that the candidate could have been discovered with fewer compounds synthesised or experimental studies. More powerful examples are prospective applications that demonstrate substantially faster than industry norms. An example is the Bradshaw project at GSK.
Can AI help to discover better drug candidates?
A recent review by Boston Consulting Group explored the results for clinical candidates from what the authors describe as ‘AI-native’ biotechs. Although the number of candidates is small so far, the results provide an intriguing first glimpse into the success rates of clinical candidates identified with the support of AI.
The authors find that 80-90% of candidates from these AI-native companies succeeded in Phase I clinical trials. This is higher than the industry standard of 40-60%. The authors suggest this is likely due to better properties of clinical candidates derived with AI-guided optimisation.
Only ten candidates from these companies have reported results from Phase II trials. Four were successful, suggesting a similar failure rate to the industry norm. If this rate is sustained, it would suggest that AI methods are less effective at reducing the biological risk of translating target activity to clinical efficacy.
We also need to be careful of hype here. Derek Lowe, in his ‘In the Pipeline’ blog, noted that of the 24 targets in the paper that were claimed to be ‘AI discovered’, 23 of them had precedence in the literature for the therapeutic indication they were targeting!
In summary
While still early, the evidence for potential cost savings and improved quality of candidate drugs discovered with the support of AI appears strong. However, the AI-native biotechs referred to in the BCG paper and GSK’s BRADSHAW benefitted from an unusually high number of computational experts supporting these projects. This doesn’t scale across the broader pharma and biotech industries. So, how do we reap the full benefit of AI in drug discovery? It is essential to make these sophisticated methods more readily accessible to experimental scientists.
It is also notable that, despite expectations of greater efficiency and quality, there has recently been an array of layoffs and restructuring among the AI-native biotechs. This may partly be due to funding constraints, as VCs move from the ‘peak inflated expectations to the ‘valley of disillusionment’ in the Gartner hype cycle. But it also illustrates that any technology platform can address only a subset of drug discovery risks. Even with AI, it remains an uncertain business.
Nonetheless, according to Amara’s law, “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.” There’s no reason to expect anything different for AI’s ultimate impact on drug discovery.
About the author
Matthew Segall, PhD
CEO and Company Director
Matt holds a Master’s in Computation from the University of Oxford and a PhD in Theoretical Physics from the University of Cambridge. He led teams developing predictive ADME models and advanced decision-support tools for drug discovery at Camitro (UK), ArQule Inc., and Inpharmatica. In 2006, he took charge of Inpharmatica’s ADME business, overseeing experimental services and the StarDrop software platform. After Inpharmatica’s acquisition, he became Senior Director of BioFocus DPI’s ADMET division and, in 2009, led a management buyout to establish Optibrium.
More AI in drug discovery resources
Augmenting Inspiration with generative chemistry
Out now in International Biopharmaceutical Industry, Optibrium’s CEO, Dr Matt Segall introduces the concept of augmented intelligence. He explains how to use dynamic…
