Seeing the power of AI in drug development

by: Andrew A. Radin

Today we announced our collaboration with Santen, a world leader in the development of innovative ophthalmology treatments. Scientists at twoXAR will use our proprietary computational drug discovery platform to discover, screen and prioritize novel drug candidates with potential application in glaucoma. Santen will then develop and commercialize drug candidates arising from the collaboration. This collaboration is an exciting example of how artificial intelligence-driven approaches can move beyond supporting existing hypotheses and lead the discovery of new drugs. Combining twoXAR’s unique capabilities with Santen’s experience in ophthalmic product development and commercialization provides a foundation for rapid discovery and development of new and improved therapeutic candidates to treat glaucoma.

A. Eye.-driven Drug Discovery

Artificial intelligence companies can be more than just service companies

We’ve had many conversations with management teams at Fortune 500 pharma companies over the last couple of years. The common message has invariably been: “We recognize that AI is important to pharma and that pharma has lagged behind many other industries in incorporating AI into the development process. But we don’t quite know how to get started or how best to leverage the capabilities of AI.”

I’m pleased that the power of artificial intelligence is being realized by pharma companies, as now is the time to apply AI techniques across the drug discovery and development process. Scientists are already using AI to design new molecules and perform predictive analytics to support clinical trials. But it is twoXAR that is taking AI right to the earliest discovery stage: shaping and validating hypotheses on targets, compounds and more. Our collaboration with Santen is proof that AI can be more than a complementary service and that AI-driven discovery approaches are poised to take a crucial seat at the figurative decision-making table.

Artificial intelligence algorithms for big data analysis, not molecular modeling.

When people think about computation in drug discovery, they often think about molecular modeling. Molecular modeling techniques typically start with a known drug target. They then use computationally-heavy physics-based modeling methods to represent both target and drug candidates and predict their interactions. Molecular modeling can also help scientists arrive at new molecular entities and has been an integral part of many drug discovery projects (Nimbus serves as a recent success story) — but can take a number of years, millions of dollars, and is limited in scale.

Our platform does not use molecular modeling techniques. Instead, it uses twoXAR-developed AI-based algorithms trained on large and diverse sets of real world biomedical data about diseases and drugs to predict which molecules might be most effective. These biomedical data include gene expression measurements, protein interaction networks, and clinical records. By examining billions of points of information, our technology is able to determine what is relevant, and what is noise, leading us to a set of associations indicating which drug might be most effective. Each of these drug predictions is accompanied with the underlying biomedical data that led to the efficacy prediction. This enables our researchers to understand possible proteins and pathways the drug is targeting.

During 2016, we leveraged our platform to identify novel treatments in four different diseases which were validated through in vitro and in vivo proof of concept studies. In each of those projects, we moved from ideas to results in a few months rather than in a few years.

How does this turn into drug candidates?

To identify novel drug candidates for glaucoma, we are using our platform to screen a massive catalog of molecules. Each drug in this catalog is associated with data (such as structure and binding affinities). These data are then linked to molecular changes in glaucoma to draw out unique disease-drug associations. Beginning with a set of molecules with known properties allows us to use them as tools to quickly confirm efficacy predictions in lab studies.

When discussing the use of known molecules as starting points, we’re frequently asked, “So, how do you protect IP?” We can take a multitude of approaches, from creating prodrugs to new formulations to screening novel catalogs of molecules for which we have IP rights. Our strategy offers two-fold benefits: a route to new IP rights to develop and commercialize a drug quickly while maximizing the opportunity for the “best molecule for the job,” regardless of its current IP status, to be brought to market.

Where do we go from here?

In the twoXAR-Santen collaboration, our goal is to find, test and bring novel glaucoma therapies to the clinic. We will use our platform to identify drug candidates and Santen will run proof of concept studies to select leads. From there, Santen can do what they do best: develop and commercialize drug candidates. We are looking forward to working with Santen. We’ve just loaded disease and molecule data into our platform and our technology has already provided new drug predictions to explore.