search

geneontology / helpdesk

The Gene Ontology Helpdesk
http://help.geneontology.org
16 stars 6 forks source link

Different annotation date for the same annotation #497

Open anphan0828 opened 1 month ago

anphan0828 commented 1 month ago

Please describe your question, idea, or concern. Dear GO team,

I have a question about the 'Date' field in GAF files. While looking through human GAF files (goa_human.gaf downloaded from GO Data Archive, https://release.geneontology.org/index.html) of June 2024, Dec 2022 and Oct 2021 releases, I realized that some protein binding (GO:0005515) annotations of some proteins are similar (same ProteinID, same GO term, same evidence, same PMID reference, same binding partner) but the 'Date' field is updated with every version. For example,

(entry in most recent version 2024-06-17 shows annotation date of 20240406) UniProtKB P61073 CXCR4 enables GO:0005515 PMID:24094005 IPI UniProtKB:Q04900 F C-X-C chemokine receptor type 4 CXCR4 protein taxon:9606 20240406 IntAct (entry in 2022-12-04 file shows annotation date of 20221112) UniProtKB P61073 CXCR4 enables GO:0005515 PMID:24094005 IPI UniProtKB:Q04900 F C-X-C chemokine receptor type 4 CXCR4 protein taxon:9606 20221112 IntAct (entry in 2021-10-26 file shows annotation date of 20210612) UniProtKB P61073 CXCR4 enables GO:0005515 PMID:24094005 IPI UniProtKB:Q04900 F C-X-C chemokine receptor type 4 CXCR4 protein taxon:9606 20210612 IntAct

I wonder why is the annotation date was updated in the new version if every other field is the same. Is this the case with only GO:0005515 protein binding annotations or can happen to other GO terms as well? Maybe there is a GO rule about updating the creation date that I did not know of. Not every GO creation date is updated with every new version.

I appreciate your help. Thank you in advance.

suzialeksander commented 1 month ago

@anphan0828 Thanks for this question. There's no automatic check/change at GO that would update the date of these manual annotations, so I would read these as curators at IntAct are reviewing/confirming existing annotations as part of their workflow.

If these were computational annotations, the changing date would indicate the pipeline to generate those annotations was rerun.

@sandraorchard can you confirm?

anphan0828 commented 1 month ago

Thank you, I took a deeper dive into this and found that annotations sourced from IntAct indeed all have the same annotation date (closest date to the GO version date) and the annotation dates get updated with every GO version. Furthermore, IntAct only annotates GO:0005515 and GO:0042802 (at least in Human) so I can filter these out quite easily if needed.

Thanks again for your help!