All of this, while CancerFoundation is trained on much less data and the model is significantly smaller.

The gene embedding learned by CancerFoundation shows grouping of genes by previously discovered Meta-programs.

CancerFoundation's zero-short cell embeddings resemble known transcriptional states established transcriptional states observed in glioblastoma (GBM) and lung adenocarcinoma (LUAD).

CancerFoundation outperformed scFoundation and DeepCDR on drug response prediction.

📢If you are interested in single-cell foundation models (scFMs), stop by our poster (West 109) at the AiDrugX Workshop at Neurips 2024. We will present CancerFoundation, a scFM tailored for studying cancer biology🧬.
Preprint: biorxiv.org/content/10.1...