Managing invalid PMID used in non-pubmed sources

[kermitt2]

http://localhost:8080/service/lookup?pmid=16262981 results in 404

However it is the PMID indicated by this record, so it exists !
http://localhost:8080/service/lookup?pii=S0266462305050762

[lfoppiano]

it looks not in the PMID source file:

(base) Johan:consolidationData lfoppiano$ zgrep 16262981 PMID_PMCID_DOI.csv.gz 
(base) Johan:consolidationData lfoppiano$ 

[kermitt2]

but present in the ISTEX ids file

lopez@work:/mnt/data/biblio$ zgrep 16262981 ~/biblio-glutton/data/istex/istexIds.all.gz
{"corpusName":"cambridge","istexId":"CC91E0F1789978CE79D653533100BA315CA337B3","ark":["ark:/67375/6GQ-9RTTRBZ7-G"],"doi":["10.1017/S0266462305050762"],"pmid":["16262981"],"pii":["S0266462305050762"]}

So it might be present in the ISTEX metadata as provided by the publishers, but not in the mapping file DOI/PMID. It is actually not a valid PMID -> https://www.ncbi.nlm.nih.gov/pubmed/16262981

This DOI does not correspond to a real article but to an index of a Cambridge journal.

I think it's fine not to map it to anything indeed with the lookup service, as it's not a valid PMID anymore. But I am not so sure how to deal with it in the full record.

[lfoppiano]

right... one thing I notice is that we do not use the pmid information from the istex mapping but only from the pmid file (which should be the authority here, isn't it?).
We should complement the pmid lookup with information from istex mapping, perhaps?

[kermitt2]

Yes I think it makes sense to complement like that the pmid lookup.

The main issue is then that the publisher metadata are not well maintained. When a PMID becomes invalid at PubMed (which is the reference), apparently they are not removed from the publisher metadata and we would have invalid mapping via the istex data file.

The submodule pubmed-glutton map DOI/PMID to MESH classes using the officiel PubMed metadata dump files. One approach to control the pmid information would be to exploit this additional mapping. I still need then to work a bit on it again.

So I would suggest to wait until I review the pubmed-glutton submodule. This module will produce an additional mapping PMID/MESH classes to enrich the record of biblio-glutton, and we could use it to control valid PMID.