Cell2Sentence / C2S-Scale (Yale, Google)
Turns a cell's gene expression into a sentence so a normal language model can reason about it; a 27-billion-parameter version proposed a cancer immunotherapy idea that was confirmed in the lab.
C2S-Scale (2025) scaled Gemma-based models to 27B parameters on cell sentences; a prediction that silmitasertib (CK2 inhibitor) boosts antigen presentation under low interferon was validated in vitro, an early example of an AI-generated, wet-lab-confirmed hypothesis.
How it works
Rank-ordered gene names as text; standard LLM training and prompting.
- Natural-language interface
- Demonstrated novel hypothesis
- Text tokenisation is lossy
- Single validated hypothesis so far
Latest papers
topQuery for this technology: (TITLE:"Cell2Sentence / C2S-Scale" OR ABSTRACT:"Cell2Sentence / C2S-Scale" OR TITLE:"Yale, Google" OR ABSTRACT:"Yale, Google") AND (cancer OR tumor OR tumour OR oncology OR carcinoma OR lymphoma OR leukemia OR leukaemia OR myeloma OR sarcoma OR melanoma OR glioma). Results are unfiltered search hits about Cell2Sentence / C2S-Scale (Yale, Google), not a curated reading list.