New Nature Biotechnology Article Highlights the Need for Cell-Cell Interaction Data in AI Biology

A new perspective article in Nature Biotechnology published on July 2, 2026 stresses that cell-cell interaction data is an essential missing layer for understanding complex biology and advancing AI foundation models in biology.

The article, authored by researchers from UCLA and collaborating institutions, argues that while single-cell sequencing and spatial biology have improved our ability to profile cellular identity and organization, they fail to fully capture how cells influence one another through contact, signaling, competition, and cooperation.

It proposes the Billion Cell×Cell Project, an ambitious effort to systematically characterize cell–cell dyads across diverse cell types and conditions, treating cell–cell interaction data as a form of functional ground truth to benchmark and improve AI models in biology.

The paper highlights Nanovials technology—which enables controlled pairing and study of individual cell–cell interactions—as a key enabler of such large-scale interaction datasets.

Results from a preprint using Nanovials to study immune–tumor cell pairs show that current single-cell foundation models do not fully recapitulate measured interaction data, suggesting there is substantial untapped value in interaction-resolved datasets for training and validating next‑generation virtual cell and tissue models.

The article concludes that integrating rich, interaction-resolved data will move AI biology beyond static cellular snapshots toward models that can predict how biological systems behave, respond, and change in context.