cartapa-dataset-checker

To validate spatial proteomics h5ad datasets for CartaPA, run the provided Python script to check structural integrity, required obs columns, protein marker counts, and CartaPA embeddings. It generates a concise validation report to prevent model misconfiguration.

The CartaPA checker supports h5ad files for CODEX-HCC, CODEX-TNBC, IMC-TNB, and SAFE-HNSCC spatial proteomics datasets. It validates basic structure, spatial coordinates, and celltype annotations specific to these multi-slice experiments.

Yes, you can check celltype annotations and spatial coordinates across multiple tissue slices in anndata objects. The checker validates coordinate integrity, celltype diversity, and treatment labels while applying dataset-specific caveats for multi-slice experiments.

CartaPA model misconfiguration occurs when spatial proteomics data lacks required obs columns, proper marker counts, or CartaPA embeddings. The checker identifies these structural and annotation issues, providing warnings for dataset-specific caveats to prevent downstream errors.

Yes, you need anndata and matplotlib installed to generate visual quality checks for h5ad spatial data. The checker relies on these dependencies alongside numpy and pandas to produce optional coordinate visualizations and validation reports.