An illustration of link formation / Courtesy of Verseon
Version is changing the drug development process, moving beyond artificial intelligence, machine learning, and high-throughput combinatorial chemistry methods to design drugs atom-by-atom. This physics-based approach enables more perfect binding and, importantly, opens up a large number of new chemical moieties for drug discovery.
The usual methods of drug discovery “involve trial and error,” Adityo Prakash, CEO and co-founder of Verseon, told BioSpace. “Even commonly used AI screenings rely on known data.” This results in the use of a tiny percentage of all possible molecules.
For example, he pointed out, “People have estimated the number of really new chemical moieties at ten33, yet the industry has produced less than 10 million chemical skeletons. Even though the collection of all combinatorial libraries can list hundreds of millions of compounds, these libraries (developed by drug suppliers and developers) have huge overlaps and contain molecules with high degrees of chemical similarity. This undermines the ability of high-throughput screening and AI trained on experimental results to find truly novel compounds. The results, he said, are “me too compounds with small variations. It’s like fishing in a drop of water compared to the whole ocean!
Verseon, however, fishes outside of this tidal pool into the largely uncharted ocean of chemical space, using molecular physics to accurately predict how new chemical structures will bind to particular proteins.
There are three steps to this process. The first is to create a virtual chemical universe made up of the rules of chemical reactions. “Others have tried to create virtual libraries,” Prakash said, but failed to consider the chemical reactions needed to make the molecules. “Verseon’s database is a dynamic molecule creation engine. We can point it to any part of the chemical universe we want,” depending on the characteristics we want.
“Creating feasible molecules and predicting synthetic steps is not enough, however,” Prakash said. “The next step is to design and test new binders for any target silicone.”
A lot silicone the work is less precise than desired, so Verseon makes the effort to understand what the binders will look like in three dimensions, how they will flex and twist around a specific chemical bond, where on the protein they will bind, how the water affects interactions and other potential problems.
“If you can model molecular physics correctly, you can run a huge array of possibilities against any protein target. To do this, we run through many chemotype options, each requiring billions of calculations, but the itself is not enough. You also need to know what trillions of calculations to perform. That’s where breakthroughs in physics come in,” Prakash said. “If you can do that, you don’t just end up with a or two results, but several hundred families with completely new chemicals that no one else would have found.”
The third step is to manufacture these molecules in the laboratory, to test them in vitro and liveand use the resulting data for AI-based optimization to obtain highly desirable pharmacokinetic and pharmacodynamic properties for the final candidates.
“If we can do all this right, the drug discovery paradigm changes,” he said. As Verseon continues this approach, it begins to fill the map of uncharted drug discovery space. Indeed, it replaces the emptiness of “behold dragons” (like a 16th century cartographer warned) with islands of known entities.
“Many companies use AI to validate their goals. It’s helpful, and we do too, but the reality is that we don’t live in a biology-poor world. We live in a biology-rich, chemistry-poor world where industry is challenged to find truly novel drugs,” Prakash said. “That’s why the R&D productivity of the pharmaceutical industry has dropped over the years.”
What sets Verseon apart is its use of molecular physics modeling to enable atomic-level engineering of new molecules. This has resulted in a drug discovery program that has 14 small molecule drugs in development: six molecules in three cancer programs, and two for liver disease, cardiovascular disease, diabetic vision loss and hereditary angioedema. “This is a completely disease-agnostic platform,” Prakash said.
The cardiovascular program is the most advanced. Verseon develops first-in-class precision oral anticoagulants (PROAC). The two compounds, VE-1902 and VE-2851, are intended for patients requiring lifelong anticoagulant-antiplatelet combination therapy. Currently, he said, “the 51 million patients in the developed world who need long-term anticoagulant and antiplatelet therapy do not have safe options.”
Preclinical and Phase I data show that VE-1902 significantly reduces blood clots, allows near normal platelet function, and allows near normal levels of bleeding. The results compared favorably to the reference drugs. The second compound, VE-2851, is chemically distinct, with a different mechanism of action. Prakash said it was more powerful than VE-1902. “It is ready for clinical trials.”
Despite the number of molecules in development, Prakash is confident that Verseon has the expertise and bandwidth to develop them. Since its creation in 2002, “we have systematically reinforced our capacities. When we decide to target a particular disease, we bring in the right experts for those programs. Over the years, we have learned to optimize our process”, to advance the molecules with maximum efficiency.
Prakash, a Caltech graduate, is listed as an inventor on more than 40 patent families. He and co-founder Eniko Fodor are responsible for developing the technology that enables all video streaming today. The third founder, David Kita, developed one of the first bioinformatics platforms and thus enabled much of the world’s genomic research for the detection of new genes and differential expression profiles. Team members who have held leadership positions in large pharmaceutical companies and scientific advisors, including Nobel laureates, have enhanced Verseon’s expertise.
Prakash believes that Verseon’s physics-based approach to novel binder design provides the foundation that enables more accurate drug characterization. Such knowledge alerts designers to look at potential problems early in the drug design process and also opens up access to a proverbial ocean of new types of chemicals that have never been made.