what is the specific dataset format?

[yingyuan830]

Hi, I'd like to use my own data to test the tool, but I am confused about the data format of MouseAtlas and not sure how to modify my data like that, could you please give some hints ? Thanks!

[sperfu]

Hi there,

The format of our input data is the npz file which you may use load_npz function from scipy.sparse package, its basically the compressed version of numpy array, which is feature by cell matirx. You may modify your input data like that.

Thanks

[yingyuan830]

Thanks. But I am still confused about the specific meaning of the features:
indices = atlas_data['indices']
indptr = atlas_data['indptr']
format_data = atlas_data['format']
shape = atlas_data['shape']
data = atlas_data['data']
Could you please give more hints so that I can modify my input data like this ?

[ycao94]

Hi, for the time being we are directly using the cell by peak matrix and vector with length n_cell to represent the data. You can modify the dataset file to implement your own data class. We'll be working on customizable data based on Anndata soon.