Codes for preprocessing of the datasets, including feature selection and sampling, unimodal representation, and calculation of the activity matrix.
Here we provide all the datasets in h5ad format. Before you start, please download and place the datasets under download/.
We provide the jupyter notebook for feature selection and sampling, python script for unimodal representation, and calculation of the activity matrix in preprocessing/ and the corresponding example shell script for running. You can process your own dataset follow the instructions.
preprocessing/scATAC_source.r was derived from Deepmaps. [Ma, A. et al. Single-cell biological network inference using a heterogeneous graph transformer. Nature Communications 14, 964 (2023).]
We use scMultiSim for simulation data generation. More details can be found here.