We founded Relation Therapeutics in 2023 to address a specific gap in computational drug discovery: the inability of existing tools to systematically connect single-cell biology to ranked target hypotheses that discovery teams can act on.
Charles brings a background in computational biology and genomics data systems, built across academic research and translational settings. He spent years building infrastructure for large-scale genomics analysis programmes before co-founding Relation Therapeutics in 2023. His work at Relation focuses on the platform architecture and the model training pipeline.
Priya is a single-cell genomics specialist who has spent years building cell-type annotation pipelines and transcriptomic analysis workflows for disease biology applications. Her expertise sits at the intersection of cell population identification and genomics data interpretation. At Relation she leads the scientific programme and drives the biological grounding of the platform.
Tom leads the engineering work behind the platform, including the data ingestion pipeline, batch harmonization logic, model serving layer, and API. He has a background in large-scale ML training pipelines and bioinformatics data engineering, with particular depth in batch-effect handling at genomics scale. He joined Relation at founding to build the infrastructure that makes the foundation model operationally useful.
Relation Therapeutics was founded in London in 2023, emerging from research conversations between Charles and Priya about why cell-type resolution was not yet systematically embedded in target discovery workflows.
The founding thesis: single-cell genomics had generated enough data and enough methodological progress that a foundation model trained on that corpus could produce biologically grounded target hypotheses, rather than statistical associations from bulk tissue that conflate cell-type composition with expression change.
We are pre-clinical and early-stage. We are building the research platform, sharing our scientific work publicly, and looking for drug discovery teams and research collaborators who want to work with cell-type-resolved target hypotheses from the ground up.
Every method and model we build is grounded in the biology. Statistical patterns that cannot be connected to a biological mechanism are deprioritized. We take the cell biology seriously even when the data science would let us skip it.
We do not consider a target hypothesis useful unless the evidence behind it is traceable. Every output the platform generates is linked to the source data that generated it. No black-box predictions.
We publish our methods, release reference datasets when data agreements permit, and share preprints before journal review. The single-cell community benefits from open science, and we want to contribute to it.