snap-stanford / UCE

UCE is a zero-shot foundation model for single-cell gene expression data
MIT License
120 stars 15 forks source link

Location of human protein embeddings #15

Closed vinettey closed 5 months ago

vinettey commented 5 months ago

Hi! I was trying to embed a human dataset using the eval_single_anndata.py. However, in file data_proc/gene_embeddings.py, the directory for protein embeddings (model_files/protein_embeddings/) does not seem to exist. Could you help pointing out where the human embedding file Homo_sapiens.GRCh38.gene_symbol_to_embedding_ESM2.pt. can be found or dowloaded? Thank you!

Yanay1 commented 5 months ago

It should be automatically downloaded and unzipped when you first run the script, but otherwise you should be able to download it from here: https://figshare.com/ndownloader/files/42715213