Relation defines the TechBio sector with AI models trans­forming drug discovery

Despite the last decade’s progress toward under­standing diseases, one stubborn problem persists. While technology now enables the measurement of cellular responses to different inter­ven­tions, the matrix of possible inter­ven­tions across proteins, contexts, and tissues is exper­i­men­tally intractable. That makes biotech R&D and even clinical testing a risky, costly process of trial and error, with failure as the norm. 

The ideal solution would be a way to do thousands or millions of experiments in silico, before a single patient is drugged, to gauge how different cell types in different disease states might respond to therapies — and, crucially, to acknowledge and incorporate time’s arrow — the harsh reality that cells and their contexts change as a disease progresses.

Relation, one of the stars of DCVC’s TechBio portfolio, has now taken on the challenge of building this simulation capability. This week the company announced it will use pertur­ba­tion data — tracking changes in live human cells in large-scale, automated lab experiments — to train a powerful new foundation model designed to predict how cells will respond to different genetic and phar­ma­co­log­ical inter­ven­tions. The model is called MORGAN, for Multi-Omic Regulatory Genomics using Artificial Neural Networks, and it will be advanced through a strategic research collab­o­ra­tion with GSK.

Under the terms of the collab­o­ra­tion agreement, announced today, Relation will generate large-scale human cellular pertur­ba­tion datasets to deepen under­standing of disease biology and support future target discovery, and may receive up to $110 million in upfront and milestone-based payments in return. That comes on top of two existing research part­ner­ships with GSK that could bring up to $108 million in upfront and success-based collaboration payments.

MORGAN will understand complex, polygenic diseases not as fixed snapshots, but as movies unfolding over time — acknowl­edging the reality that drug inter­ven­tions affect gene and regulatory pathways differently at various points in the script. Building such a model starts with real biological data from high-resolution, multi-omic studies of human disease-relevant cellular systems. And it accepts the likelihood that successful inter­ven­tions will need to be bifunc­tional or even multi-functional, hitting multiple genes and pathways at the same time. 

Building a general-purpose cellular foundation model is a hugely challenging project, and could prove to be the most audacious step forward in biological computation since the emergence of models like AlphaFold for protein structure prediction or Boltz for under­standing molecular inter­ac­tions. The team at UK-based Relation, which we’ve backed since 2021, have so far been char­ac­ter­is­ti­cally British in their modesty — and the internal programs they’ve disclosed have focused on a single disease type, osteo­porosis. (Still, a telling choice, given the challenges of treating the real causes of this hitherto intractable disease.) But with the MORGAN announce­ment, the true scale of the company’s ambitions is starting to show. And the collab­o­ra­tion agreement announced today is a striking show of confidence in the program from a pharmaceutical giant. 

Beyond MORGAN itself, I’m excited by the potential scale of Relation’s biology-first approach. The company has chosen to build systems that address the key factor that improves the probability of success in drug development: confidence in biology. By focusing on better under­standing disease biology, Relation is allocating capital with a better return across a broad portfolio — in target discovery, drug asset prior­i­ti­za­tion, and trial design — and building a platform-first TechBio company with a rich pipeline of future medicines.

Today’s announce­ment marks the beginning of the next chapter of Relation. MORGAN’s insights will change the way we think about pathology in profound and important ways — and I can’t wait to see what Relation does next. As the company reminds itself and the world, ​“The patient is waiting.”