An unofficial implementation of 'CellDART: cell type inference by domain adaptation of single-cell and spatial transcriptomics data' (link) based on pytorch-lightning.
The major requirements of establishing the environment are listed as follow.
pytorch-lightning==1.5.10
scanpy==1.8.2
pandas=1.3.4=py38h8c16a72_0
numpy=1.21.2
scipy=1.7.1
pytorch=1.8.0
You may refer to the pytorch official website for installing proper distribution with CUDA on your machine.
We also combined scvi-tools analysis within the preprocessing step.
scvi-tools=0.15.0