About the Event
The pharmaceutical industry has become notorious for being a poor return on investment. Amortized development costs for each new drug that comes to market now reaches north of $5 billion. The reason for this low productivity lies in the fact that the current drug discovery process is based on trial-and-error. Using a mixture of combinatorial chemistry and high-throughput screening, pharmaceutical companies generate large batches of chemical variants and then test for the presence of any significant protein-molecule interactions.
Although combinatorial chemistry revolutionized the pharmaceutical industry in the 1990s with its ability to generate large libraries of different molecules, it is highly limited in scope. The technology is centered around making small variations on pre-existing molecular scaffolds. Consequently, many of the new drugs that are approved by the FDA are based on already existing drugs with some side-chain modifications. Out of the possible chemical space with 1020 "drug like" molecules, companies are only screening a tiny space of roughly six million compounds.
However, exploring a larger chemical space is a hard problem. Current laboratory processes cannot scale up to accommodate synthesizing even 100 million diverse molecules. Rather than focusing on scaling up the screening process, the real answer lies in building a method that can accurately explore a vast space of never-before-synthesized chemical options against a disease-causing protein.
At Verseon, we have worked for 15 years to build a systematic, physics-driven platform for drug design. Using a computationally-driven approach, we can construct a very large chemical space of drug-like molecules that have never been synthesized before. In this talk, I will describe the challenges behind constructing such a space and actually exploring this virtual space without first synthesizing the compounds. We start with a protein target of interest, construct a large chemical space of novel molecules and then model the interactions of these molecules with the protein in an aqueous environment. The modeling of interactions presents many challenging physics problems, as simple brute force is computationally intractable. This has motivated our decade-long quest towards developing new ways of modeling the physics at appropriate scale lengths so that we can get the desired accuracy while keeping the problem computationally feasible. I will also touch on some of the drug programs we are now developing using our platform.
Questions? Contact: : email@example.com