Inquiry and question about data usage in pretraining the foundation model

[wuwusky]

Thank you for the excellent work!
I am interested in learning more about the pretraining process. Based on the paper, it appears that the scFoundation model was trained exclusively on single-cell gene expression matrices, including the count values calculated internally and the RDA training strategy. Could you clarify if any metadata or additional prior knowledge was incorporated during the pretraining process?

[WhirlFirst]

Hi, thank you for your interest! We didn't use any metadata or prior knowledge for pre-training.