Cell type matched wrongly

[Hrovatin]

When loading sfaira (human pancreas) I have noticed some cell types are matched to ontology wrongly.

Both should be "pancreatic PP cell": gamma | ['pancreatic endocrine cell'] gamma cell | ['pancreatic endocrine cell']

[davidsebfischer]

Thanks @Hrovatin, do you know which data sets in particular?

[davidsebfischer]

I found:

• [x] https://github.com/theislab/sfaira/blob/dab9cb3f6d89dc6c6b52f9ce23c9a7e938e4d91b/sfaira/data/dataloaders/loaders/d10_1016_j_cels_2016_08_011/human_pancreas_2016_indrop_baron_001.tsv#L10 (flagged here https://github.com/theislab/sfaira/blob/dab9cb3f6d89dc6c6b52f9ce23c9a7e938e4d91b/sfaira/data/dataloaders/loaders/d10_1016_j_cels_2016_08_011/human_pancreas_2016_indrop_baron_001.py#L11), need to remove class map too: https://github.com/theislab/sfaira/blob/dab9cb3f6d89dc6c6b52f9ce23c9a7e938e4d91b/sfaira/data/dataloaders/loaders/d10_1016_j_cels_2016_08_011/human_pancreas_2016_indrop_baron_001.py#L35
• [x] https://github.com/theislab/sfaira/blob/dab9cb3f6d89dc6c6b52f9ce23c9a7e938e4d91b/sfaira/data/dataloaders/loaders/d10_1016_j_cmet_2016_08_020/human_pancreas_2016_smartseq2_segerstolpe_001.tsv (flagged here https://github.com/theislab/sfaira/blob/dab9cb3f6d89dc6c6b52f9ce23c9a7e938e4d91b/sfaira/data/dataloaders/loaders/d10_1016_j_cmet_2016_08_020/human_pancreas_2016_smartseq2_segerstolpe_001.py#L10)

[davidsebfischer]

new term: "pancreatic PP cell", "CL:0002275"

[davidsebfischer]

PP is a sub term of "pancreatic endocrine cell" (https://www.ebi.ac.uk/ols/ontologies/cl/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FCL_0008024&viewMode=PreferredRoots&siblings=false), I think at the time we did not know the exact term because gamma is not annotated in cellontology. Therefore, we chose the umbrella term.

[Hrovatin]

Yes, gamma was an old name for Pp cells, but it is still often used in literature. Maybe add this into ontology matching in general as it will likely be the same in some new datasets as well.

[davidsebfischer]

I won't hard-code it to the semi-automatic matching as this is technically a synonym of the cell type label, ie something that cell ontology takes care of managing and that we simply consume here.

[davidsebfischer]

We could ask cellontology to update it though!