scMVAF is a multi-view integrated clustering framework that enhances feature diversity by fusing features from multiple perspectives to facilitate the identification of cell subtypes. The adaptive fusion of different perspectives enhances the representation learning of data, and the complementarity between perspectives helps to more comprehensively capture the diversity and complexity of data, thereby improving the robustness of clustering.
python --- 3.9.19
numpy --- 1.24.4
pandas --- 1.5.3
scanpy --- 1.10.0
scikit-learn ---1.4.1.post1
torch --- 2.0.0+cu118
torchaudio --- 2.0.1+cu118
torchvision --- 0.15.1+cu118
- data.mat is a MATLAB-formatted dataset used as input for the model.
- data['X']: Count matrix, where each row represents a cell and each column represents a feature (e.g., gene expression counts).
- data['Y']: Ground truth labels for each cell.
The datasets folder includes the preprocessed deng dataset. To use a different dataset, you need to modify the following code:
parser.add_argument('--data_name', default='deng')python scMVAF.py