Question about how the "vocab" created.

[Alex2975]

Dear Authors,

Thank you very much for making this tool public. Very impressive. I have a question about the vocab.json. The gene names are used as tokens. How did you created the vocab.json? I search for some of the gene names in the json file, and I cannot find the exact names. Could you please share how the gene names were collected for training the base model? If I have a new scRNAseq data set, how can I map my gene Ensembl IDs to the gene names in the vocab.json? Thank you again.

[subercui]

Hi, since we collected the data from CellXGene, we followed their gene names there as well. You can find a mapping between Ensembl IDs to the names in the attached file

gene_info.csv

[Alex2975]

Thank you so much for the help. I will try the csv file.