The Company That’s Applying Machine Learning to Virtual High-throughput Screening

Drug discovery is long and costly: bringing a new drug to market can take 15 years and cost over $2 billion. It is also risky: of the thousands of chemical compounds screened as drug candidates, only 10% ever make it to clinical trials (a preclinical failure rate of 90%), and of these, the overwhelming majority fail in the clinic. Thus, the pharma industry is desperately seeking innovative solutions to create better compounds and to understand why so many fails.

One area promising technology for optimizing drug discovery is artificial intelligence, especially machine learning. The French company Synsight had the idea of applying machine learning to virtual high-throughput screening (HTS).To learn more about their approach, Scientist interviewed Cyril Bauvais, co-founder and CEO of Synsight.

Cyril, tell us more about you.

I hold a Ph.D. of chemoinformatics and I am specialized in computer-aided drug discovery and design, more particularly in high-throughput virtual screening (virtual HTS) and artificial intelligence based chemoinformatics.

About your lab, what is the history, its mission?

Synsight is a Paris-based SME working in computer-aided drug discovery (CADD).I co-founded the company in 2013 with Guillaume Bollot. We have a dual business model: as an R&D partner/service provider and as a drug discovery company with our own pipeline and are, for now, self-financed by our CRO activity.

A third full-time scientist, Pierrick Craveur, our team. The three of us, all PhDs, have complementary expertise and several years of research experience: Cyril, in chemoinformatics and biological macromolecule modeling; Guillaume, in protein and enzyme design and in quantum chemistry; and Pierrick, in computational structural biology.

Using our proprietary CADD platform, we aim to make drug discovery faster, cheaper and smarter, by providing our partners with novel scientific insights and helping them to make better decisions. We also run our own drug discovery research programs in areas such as RNA splicing and immuno-oncology.

What services does your lab offer?

We primarily work with biologists and chemists at pharma and biotech companies as well as with academic researchers. We help them across the whole drug discovery value chain: target identification, hit finding, lead discovery & optimization, ADME-tox, drug repurposing, etc. Although our focus is early-stage drug discovery, our CADD platform is amenable to other life sciences and physical sciences domains: for example, agrochemicals or materials science.

Basically, if you can draw a molecule, then we can model, screen and study it!

What sets your expertise apart from others?

Early-stage drug discovery includes many design-and-test cycles. A large Pharma company may go through 25 cycles or more in a single year! This is where Synsight comes in: we help our clients and partners to make focused decisions on :

• which molecules to make;
• how and why certain molecules exert a biological activity (or fail to);
• how to enhance the activity or selectivity of lead molecules, or make them more drug-like;
• and even how to find new indications for known drugs.

Importantly, CADD does not replace experimental chemistry and biology, but it does add layers of complementary information to help chemists and biologists design and run experiments. Indeed, by marrying human expertise to the power of machine learning and other computational approaches, you can get the best of both worlds. The knowledge we generate can even help outside of the laboratory: for instance, by providing scientific arguments to inform the patent strategy for a new drug candidate.

Here are two recent testimonials:

“By providing us with strong expertise in Modeling and Organic Chemistry,_Synsight _delivered clear answers to our research questions. The in-silico application generated in our collaboration enabled us to accelerate screening of our compound library, such that we could narrow down the number of compounds to test in vitro.”
- Medicinal Chemist, Head of R&D at a European SME

“_Synsight _clearly identified the enzymes responsible for the pathology that we are studying. They also identified a set of inhibitors to selectively target those proteins.”
- Biologist Head of Research at a French Public R&D Institute

What challenges are you willing to address in the near future?

We have been working diligently to make our CADD platform even more powerful! For example, we recently expanded our Virtual Screening library from tens of millions of compounds to hundreds of millions, with a concomitant amplification in chemical diversity. We are also adding modeling functions to acquire new types of information in areas such as Virtual Med Chem.

Although we understand some of the skepticism about artificial intelligence (AI) in drug discovery, over the past few years we have judiciously been incorporating into our platform specific machine learning and deep learning algorithms developed in academia. We use these only where we think they can add real value, like in multi-parametric analysis.

Further expansion of our team will be partially contingent on the funding and investments that we acquire for our own drug discovery projects.

What do you expect from Scientist?

I would mainly expect to meet and discuss how our CADD platform could help researchers advance drug discovery or other R&D projects.

If you are interested in knowing more about Synsight’s offer, visit its profile.