load_gptcelltype

datasets.load_gptcelltype()

Load the bundled GPTCelltype (Hou and Ji 2024) annotation agreement scores.

Reads gptcelltype2024_agreement.csv (one row per cell type per method that was run) and reduces it to a method by dataset by metric tensor: per (method, dataset) the mean agreement over the cell types and the fraction of cell types that fully match the manual annotation. Datasets and their features come from gptcelltype2024_features.csv, in its row order. A method not run on a dataset has no rows there and surfaces as numpy.nan; nothing is imputed or dropped.

Returns

Type	Description
GPTCelltype