GSK and Relation Therapeutics have entered a $110 million research collaboration to generate biological datasets for training AI models in drug discovery.
Key facts
- •The GSK-Relation collaboration is valued at up to $110 million.
- •Relation’s Lab-in-the-Loop approach integrates laboratory experimentation with computational analysis.
- •CZ CELLxGENE provides access to more than 100 million standardized cells.
- •A Nature Methods study found that increasing training data did not consistently lead to performance gains in single-cell foundation models.
- •Relation is currently using its proprietary Osteomics platform to investigate osteoporosis.
GSK has entered a research collaboration with the British biotechnology firm Relation Therapeutics worth up to $110 million. The agreement expands the companies' existing work in AI-assisted drug discovery by combining biological data generation with AI model development.
By the numbers
Data Generation and AI Development
Under the new agreement, Relation will generate large-scale datasets measuring human cell responses to genetic changes and drug interventions. This data will train AI models, including those within Relation’s MORGAN platform, to identify potential drug targets. The collaboration builds on previous work between the two companies focused on osteoarthritis and fibrotic diseases.
Challenges in Biological Data Training
Recent research highlights the complexities of using biological data for AI. A 2025 review in Experimental & Molecular Medicine noted that public repositories like CZ CELLxGENE contain vast amounts of data, but technical noise, batch effects, and dataset redundancy can hinder model performance. Additionally, a study in Nature Methods found that single-cell foundation models do not always follow clear data-scaling laws, suggesting that simply increasing training data does not consistently improve results.
Trends in AI-Biopharma Partnerships
A 2025 analysis in Nature Biotechnology identified specialized dataset providers as a key trend in biopharma partnerships. Other recent industry deals include GSK’s $37.5 million agreement with Ochre Bio for liver data and a $200 million deal between AstraZeneca, Pathos AI, and Tempus. These partnerships reflect a broader industry focus on securing high-quality, disease-specific data to overcome bottlenecks in AI-driven drug discovery.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by Artificial Intelligence News.
