Data preprocessing

[2001229323jyf]

Hello, it is great work!

However, I encountered difficulties when attempting to convert the dataset I found into the .h5ad format and integrating it into your project. I observed that you provided preprocessing methods for the data in the project as follows:

1.Exclude spots with less than 500 UMIs and genes expressed in less than 3 spots
2.Normalize the expression matrix with the LogNormalize method in Seurat.
3.Annotate the cell types by label transfer (the TransferData function in Seurat) with single-cell breast cancer dataset GSE118390 as reference dataset.
4.Deconvolution results stored at adata.obsm['predicted_cell_type']
Cell-type label (the max value) stored at adata.obs.cell_type

If you could provide some code guidance or suggestions on data preprocessing in these aspects, or if you could offer more well-processed data in the .h5ad format, I would be greatly appreciative.

[lhc17]

First of all, thank you for using it~ My preprocessing steps are following the standard process of Seurat and Scanpy. Can you tell me exactly which step you encountered an error after converting the dataset to h5ad, and can you perhaps provide me with the structure of your current ‘adata’?