Pretrain with scATAC data

[WhenMelancholy]

Hello, it is mentioned in the paper that scGPT can utilize multi-omics data. May I ask if it is possible to use scATAC data for pre-training? Or how can such related data be used for fine-tuning downstream tasks?

[fscdc]

Hello, it is mentioned in the paper that scGPT can utilize multi-omics data. May I ask if it is possible to use scATAC data for pre-training? Or how can such related data be used for fine-tuning downstream tasks?

I face related problem, I want to pre-train a new foundation model with my own datasets using scGPT model architecture, but I do not know the method