exact canonical match of resolver includes NCBI subspecies/strains

[jhpoelen]

when using resolver.globalnames.org with "Homo sapiens", exact canonical matches include both NCBI:9606 (Homo sapiens) and NCBI:741158 (Homo sapiens ssp. Denisova). Is this expected?

[dimus]

It is, unfortunately, because susbspecies epithet is capitalized (which is against zoological code rules as far as I know). As a result "Homo sapiens ssp. Denisova" is parsed as an 'unknown' subspecies of species Homo sapiens described by Denisova with canonical form detemined as Homo sapiens

[jhpoelen]

Any way we can convince NCBI to update their capitalization and conform to the zoological code? Do you know anyone on their taxonomy team?

[dimus]

another problem is that I do not think anybody officially described Homo sapiens denisova so in a way the result you got is probably correct, as at the moment it should be placed as the same species

[dimus]

I do not think NCBI pretends to be taxonomically correct, they have all kind of things in their data

[jhpoelen]

Ok, would also explain matches like those observed in http://www.globalbioticinteractions.org/?sourceTaxon=Phytophthora%20infestans&interactionType=interactsWith , where potato late blight fungus
(Phytophthora infestans) is resolved to https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=403677 (Phytophthora infestans T30-4, a strain) and https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=4787 (Phytophthora infestans, the actual species)?

[dimus]

Hm, this is actually bad, we need to penalize matching score for situations like this

[jhpoelen]

There's a bunch of this happening for pathogen-host interactions. I updated the title of this issue to include "strains". Please let me know if you need more examples or how else I can help resolve this.

[dimus]

I added a ticket to penalize score of matching results if parsing was not 'clean'. In both cases the name strings would be marked as "parsed with mistakes" by gnparser GlobalNamesArchitecture/gnparser#291

[jhpoelen]

Great! Closing this issue as duplicate of GlobalNamesArchitecture/gnparser#291 .

[jhpoelen]

Re-opening issue. It appears that the issue still persists in the https://resolver.globalnames.org , where a search for "Procladius sp1 M_PL_014" results in exact canonical matches to fly strains (NCBI:1981569, NCBI:1981570, NCBI:198157, NCBI:1981572, NCBI:1981573, NCBI:1981574) as well as including the expected canonical match to the fly genus NCBI:191633 .