I founded Magnetic Ventures in 2018 with the fundamental investment thesis that we are at the dawn of the second golden age of healthcare, driven by the convergence of technology and biology. We believe technology tools including artificial intelligence (AI) and machine learning (ML), combined with omics are enabling us to understand the biology of disease in ways previously unimaginable and have the potential to transform drug discovery and development.
It’s estimated that a novel therapeutic requires a minimum of $1B of funding (The Tufts Center for the Study of Drug Development estimates this number is closer to $2.5B) and 10+ years of development to achieve FDA approval. Despite this investment, less than 10% of targets entering clinical trials ultimately receive FDA approval.
This is the problem Relation Therapeutics is on a mission to solve.
Relation combines patient tissue-derived atlases of disease at single-cell resolution, functional genomics, and graph ML to prioritize interventions with the greatest likelihood of combating disease mechanisms. Drawing on advances in experimental feedback systems, their “Lab-in-the-Loop” enables the ML to make predictions from complex datasets in order to parse the most important information for understanding the problem itself.
Every year, the resolution of high-dimensional molecular probing (omics) to the human body and the amount of data generated are exponentially increasing. In parallel is the accelerated development of techniques required to learn meaningful representations from complex data (AI, deep learning, ML). Applying these techniques to integrated multi-modal biomedical data leads to an unprecedented understanding of human disease biology.
Relation’s platform applies a specific type of ML, graph ML, a technology best known for its application to recommendation systems used by companies such as Twitter, Netflix, and Google Earth. Everyone has seen a version of “since you bought/watched/liked this, you may want to consider buying/watching/liking these items next” — why not apply this to drug discovery? These recommendation systems are successful in other domains for two primary reasons: 1) the amount of data, and 2) representation learning on graphs. Now imagine you apply this insight to biology: “since you have already successfully tested these molecules/targets, you may want to try these molecules/targets next.” Relation has the tools in place to understand a previously insurmountable number of combinatorial functional relationships between genes, proteins, and therapeutic compounds, which we believe will dramatically increase the probability of success in drug discovery.
As early-stage investors, we make bets not only on category-defining technological innovation, but also, and equally as important, on teams. When a founder we know and respect calls with an idea we pay attention. Relation co-founder and Board Chair Charlie Roberts is someone I have known and deeply respected for close to 15 years. When Charlie called about Relation, I jumped at the opportunity to learn more.
It is extremely hard to find a team that rivals Relation’s expertise in mechanistic biology, drug discovery, and proficiency in ML. The team is led by CEO Dr. David Roblin: the former head of R&D and CMO of Pfizer Europe and includes CTO Dr. Lindsay Edwards: the former VP of AI at AstraZeneca and VP of ML at GSK. In addition, the Scientific Advisory Board is world class from institutions including Stanford, The Broad Institute of MIT and Harvard, Imperial College of London, and Oxford.
We believe Relation has the ideal recipe of technology and team to become a pioneer in data-driven drug discovery. We are proud to be partners to the team as they execute their mission to decode and modulate biology to cure disease, faster and with greater certainty, and to transform the discovery and development of new medicines.