What Artificial Intelligence Brings To Drug Discovery – Sanofi Interview
Executive Summary
By adding another AI deal to Sanofi's innovation quiver, the French group says it's widening its hunt to find single molecules that offer dual or multi-specific targeting to fight disease.
Sanofi says it aims increasingly to employ artificial intelligence in its quest to find single molecules which can hit more than one disease target, thereby increasing the efficacy of future novel medicines.
Its latest collaboration – announced May 9 with AI specialist Exscientia of Scotland – won't be the company's last in that space, either, Sanofi's external innovation chief says.
Outlining the French drug maker's thinking in an interview with Scrip, Adam Keeney noted that Sanofi has been exploring use of artificial intelligence in drug discovery for quite some time, enticed by the promise of combining Big Data and AI.
"We have tested a number of different technologies and have begun a number of collaborations to think through how we approach the inevitable breakthroughs that are about to occur with this sort of technology," Keeney said. One is with IBM which uses the Watson question answering computer system to drive some of Sanofi's drug discovery efforts.
Another strategic collaboration is with Schrödinger, Inc, using that New York-based group's computer-aided drug design process. Begun two years ago, that arrangement has Schrödinger providing target analysis and validation upwards to lead asset identification and lead optimization for up to ten drug discovery programs. Keeney said that process has now been integrated into many of Sanofi's own discovery projects.
"The arrangement with Exscientia is a simar sort of collaboration - to explore technology and how it will impact our drug discovery process. This is part of an ongoing evolution and evaluation of how to keep on the cutting edge, to aid our efforts in drug discovery," Keeney said.
Sanofi external innovation head Adam Keeney
Sanofi
The Exscientia AI Model
Explaining the logic behind the multi-target approach using AI, Keeney noted that "it's unlikely you're going to be able to take a disease as complex as diabetes and try and assume a single target can take care of all of the symptoms and maladies of that disease. So we've been exploring single molecules that have dual or multi-specific targeting so that you're able to take advantage of the full pharmacology with one molecule."
He said Sanofi's newly approved biologic for atopic dermatitis, Dupixent (dupilumab), is a case in point.
"That's an example of a biologic that is actually working through two independent mechanisms; interleukin-4 and interleukin-13 inhibitors – and that's showing tremendous clinical benefit. The Exscientia opportunity provides us a look into poly-pharmacology, but with a small molecule approach."
"If we could synthetize one small molecule that could work on two targets at once, that is a great opportunity to treat these complex diseases - so, if the technology demonstrates itself, we could get glucose control and weight management from one molecule which would simplify the development path and the regulatory path, because it's one molecule rather than combining several separate chemistries," he said.
The collaboration with Exscientia will also focus on co-morbidities of diabetes patients. "We're particularly interested in lipid management, weight management, and conditions such as NASH, or chronic inflammation of the liver. We're thinking of applying this technology to the suite of metabolic diseases; that includes diabetes but is not limited to that, and includes cardiovascular and liver effects of living with high weight and high glucose."
Research Phase Funded By Sanofi
Sanofi's AI drug discovery project with Exscientia has a research phase that Sanofi is paying for.
"There we can explore a number of targets in the metabolic space, and if that is successful, then the collaboration can be extended and Sanofi can take the molecules further in preclinical and clinical development."
"In the initial phase, it's basically an Exscientia project. They deliver data for us to review and we would then have an opportunity to take that program further forward and then the new chemistry and IP would be owned by Sanofi."
The project has pre-set milestones. It will initially involve Exscientia designing the algorithms based on specific targets.
"We're working on between 40 and 50 single targets, but if you think about putting those together in bispecifics, that's a tremendous amount of combinations possible," Keeney said.
Explaining the process stages, he said: "there's an early stage where in silico [performed on computer or via computer simulation] you do the early screening, then you apply the algorithms that Exscientia have been working on to come up with a number of target hits."
"From there we'd take those hits and synthetize them into real compounds, and then do an intra process of improvement. There are a number of milestones along the way before we would declare what might become a lead candidate, after which Sanofi has more responsibility for the product. So, there are a number of near-term research-based milestones, and then, if we were to take a molecule further forward, there are a number of preclinical regulatory milestones as the program advances," he explained
In summary, he said Sanofi's main key interest in drug discovery going forward is "the ability to generate one single small molecule that can have pharmacological effects on multiple mechanisms could actually open up a whole new treatment approach."
That will necessitate artificial intelligence working alongside therapeutic insight and human creativity.
"That needs the power of data to be able to crunch those combinations and come up with leads because that's not an intuitive process – you need to use the power of numbers to come up with the molecules that have that pharmacology, so we have our eye on the strategic goal of having single molecules to treat multiple parts of the disease and so to give an overall better outcome for the patient," Keeney concluded.