h5ad Data Format Description

Input h5ad Format Description

• st.h5ad: Spatial transcriptomics expression matrix AnnData object, which is required to contain the raw expression matrix, gene names, spatial coordinates, and other information. For details, please refer to SDAS Input Data Format Description. For multiple slices of spatial transcriptomics data, you can provide a custom slice_key (default is 'sampleID'), which is used for plotting (and for providing batch information in cell2location). If not provided, the data will be treated as a single slice.

• sc.h5ad: Single-cell expression matrix AnnData object, which is required to contain the raw expression matrix, gene names, and other information. For details, please refer to SDAS Input Data Format Description. At the same time, the single-cell obs must contain annotation information, stored in a custom label_key. The RCTD algorithm does not support annotation information containing '/'.

• The SCimilarity algorithm is only applicable to human samples.

Output h5ad Format Description

• Input h5ad + annotation results: The score for each cell type is stored in obsm['anno_score_<algorithm_name>'], and the type with the highest score is stored in obs['anno_<algorithm_name>']. Other output results and intermediate files of the algorithm are not saved.

• Cells that are not annotated are uniformly labeled as NA.