90% of drugs that enter clinical trials fail, and each failure costs on average $97 Million. Drug R&D is too risky and too expensive, because we develop drugs against the wrong target or with unacceptable side effects, test them in a patient population that is too broad or a poor match to the drug, and defer failure until late in the game. We use technology to de-risk development, avoid wasted effort, and feed lead molecules more cheaply into the R&D pipeline.
We combine molecular profiling (sequencing and measuring the levels of molecules in cells) with our maps of biological networks to generate network models that describe compactly how cellular biology changes in disease and upon exposure to drug. Our models serve as ideal foundations for machine learning, that we apply to de-risk and reduce cost across the R&D cycle.
Drug and disease mechanisms are rarely local; their effects propagate broadly across cellular networks and patients differ in how their networks propagate these effects. Current analyses of molecular profiles apply statistics to find individual measures that vary and then link them to biology using databases that only put half of human genes in biological context. By contrast, we first project these data onto our network maps that cover 90% of human genes and then analyze them at the network level for better signal-to-noise and deeper insights. Our advantage lies in the breadth and quality of our network maps and databases of disease and drug molecular profiles.
By comparing disease and drug network models, we gauge whether a drug selectively reverts the patterns and processes of disease. As a secondary filter on the outputs of high-throughput screening, we improve the quality of hits. The same method applied to data with perturbations of individual network genes guides rational drug design. The benefits are particularly strong for complex diseases lacking validated targets or reliable functional assay readouts.
By measuring cellular response as drug structure and dose varies (efficacy and side effect risk), we guide lead optimization. We make preclinical studies more reliable by ensuring network features are conserved between human patients, in vitro assays, and animal models. We pre-identify patient subgroups on their molecular match to drug to select and stratify them in trials. We integrate molecular profiling into trials to build Bayesian predictive models of drug response that optimize later-stage trials and inform go/no-go decisions. We feed seamlessly into health economics and outcomes research and enable individualized therapy in clinical practice.
We focus on complex, poorly-understood diseases lacking effective treatments where our platform delivers the greatest value. Our initial efforts are directed at oncology, degenerative and neuro-degenerative conditions, and rare diseases. We seek to serve and partner with pharmaceutical companies, and others dedicated to delivering effective therapies, to develop our existing assets, those of our partners, and those we discover jointly.
Serial Entrepreneur in Healthcare Technology (4 exits)
Consultant to Healthcare Industry on Innovation
MD (Yale) and HMS Faculty
Chief Executive Officer
13 Years in Drug Development and Commercialization Consulting
Chemical Engineer; Former MIT Faculty
Experienced Life Sciences Founder and CFO (2 exits at >$1B)
Founder & Managing Partner (Soar Ventures)