finding AGE from cells in scimilarity

[Hanadabdullahi]

Hi!

Ive been using scimilarity in order to find cells that are close in gene expression, using these cells thereafter to predicts an average age of the key cell compared too. But ive been struggling trying to extract an age from the cells that I get after doing a comparison. Is that value saved somewhere? or easily extractable? or is there any donor id for the cellxgene dataset that I can use? so far the sample id does not work with the cellxgene census, which means that its hard to extract the original metadata for each cell in the scimilarity.

Am I being dumb? or is there an easy solution for this?

[tony-kuo]

Hello. We didn't include age as metadata, sorry.

The cellxgene sample ID is derived from column data in downloaded h5ad files. They are gathered from these columns, in this order of priority: "sampleid", "batch", "sample_id", "sample_uuid", "donor_id", "donor_uuid", "subject", "patient", "individual", "orig.ident".

We did not use the census API so all our values are derived from the h5ad downloads. Hope this helps.
Perhaps the "study" column in cellxgene data will be useful, as they correspond to the collection UUID.

[Hanadabdullahi]

Hi! Just ensure that I understand correctly, the sample column in the metadata in the scimilarity for the cells that come from cellxgene datasets can have different type of entries. In one case it could be the batch if we dont have "sampleid" and in another case it could be the "sample_uuid" if we dont have the other data that are earlier in the priority list? mvh Hanad Den tis 4 mars 2025 kl 00:05 skrev Tony Kuo ***@***.***>:

…

Hello. We didn't include age as metadata, sorry. The cellxgene sample ID is derived from column data in downloaded h5ad files. They are gathered from these columns, in this order of priority: "sampleid", "batch", "sample_id", "sample_uuid", "donor_id", "donor_uuid", "subject", "patient", "individual", "orig.ident". We did not use the census API so all our values are derived from the h5ad downloads. Hope this helps. Perhaps the "study" column in cellxgene data will be useful, as they correspond to the collection UUID. — Reply to this email directly, view it on GitHub <#32 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A3LKZ5MCHNO6KF6UWVYFQZT2STN5PAVCNFSM6AAAAABYHZ2RD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOJVG43DQMRSGU> . You are receiving this because you authored the thread.Message ID: ***@***.***> [image: tony-kuo]*tony-kuo* left a comment (Genentech/scimilarity#32) <#32 (comment)> Hello. We didn't include age as metadata, sorry. The cellxgene sample ID is derived from column data in downloaded h5ad files. They are gathered from these columns, in this order of priority: "sampleid", "batch", "sample_id", "sample_uuid", "donor_id", "donor_uuid", "subject", "patient", "individual", "orig.ident". We did not use the census API so all our values are derived from the h5ad downloads. Hope this helps. Perhaps the "study" column in cellxgene data will be useful, as they correspond to the collection UUID. — Reply to this email directly, view it on GitHub <#32 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A3LKZ5MCHNO6KF6UWVYFQZT2STN5PAVCNFSM6AAAAABYHZ2RD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOJVG43DQMRSGU> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[tony-kuo]

Hi. That is correct, we try to get the sample ID and that is the order of columns we check to try and obtain it.