lookup_tax_data: When looking up taxonomy of binomials, only look up genus

[zachary-foster]

Often, there are many species in a data set sharing a genus. When looking up the taxonomy from taxon names, it is inefficient to use the full name in many cases since that taxonomy of many species can be inferred from a single query of the genus. Also, sometimes a genus is in a database while a species is not. However, this might cause a problem when one genus names is used for multiple genera.

> species_names
  [1] "Achlya hypogyna"                  "Albugo liabachii"                 "Aphanomyces astaci"               "Aphanomyces cladogamus"          
  [5] "Aphanomyces cochliodies"          "Aphanomyces euteiches"            "Aphanomyces frigidophilus"        "Aphanomyces invadans"            
  [9] "Aphanomyces laevis"               "Aphanomyces stellatus"            "Bremia"                           "Hyaloperonospora parasitica"     
 [13] "Peronosclerospora maydis"         "Peronosclerospora philippinensis" "Peronosclerospora sacchari"       "Peronosclerospora sorghi"        
 [17] "Peronospora belbahrii"            "Peronospora tabacina"             "Peronosppora destructor"          "Phytophthora alni"               

[sckott]

this might cause a problem when one genus names is used for multiple genera.

does this mean many genera in different higher taxonomic groups? e.g., the same genus name in plants and animals?

[zachary-foster]

Yea, like "Achlya", which is a moth and an oomycete.

[sckott]

possible to allow user to give a higher taxon group ?

[zachary-foster]

That would work well when the user is only looking at one Kingdom, which I expect is usually the case. I think that would have to be handled by taxize though, since if there are multiple matches, only one can be returned. If there was a variant on taxize::classification that returned the taxonomy of all matches, perhaps with an additional column of numbers to specify which match each rank belonged to, then that could be implemented in taxa.

[sckott]

If there was a variant on taxize::classification that returned the taxonomy of all matches, perhaps with an additional column of numbers to specify which match each rank belonged to

there is the concept in taxize::get_* functions to get all results when > 1 result - and not go through the prompt - but taxize::classification has no equivalent.

though perhaps this is close enough: use get_* to get any number of taxon IDs, then pass to classification, optionally bind classifications together

or does that not do what's needed for this?

[zachary-foster]

I think that would work. Could add an option to taxize::classification named all_matches = TRUE that returns that output format. Probably should always have that extra column if TRUE, even if only one match is returned so the format is consistent.

[sckott]

Which thing would work?

[zachary-foster]

there is the concept in taxize::get_* functions to get all results when > 1 result - and not go through the prompt - but taxize::classification has no equivalent.

Sorry for the vagueness. Looking at the code, it looks like taxize::get_* are called when a taxon name is supplied and they handle the prompting of the user. An argument could be added to classification that allows for multiple returns from taxize::get_* without a prompt, looks up the classification of each, and then rbinds them together with an extra column with the matching taxon ID.

[sckott]

Thanks for clarification.

Sounds good up to the point of rbinding together - that would mean a departure from the output format in other cases - and I'd rather not have variable output formats

opened an issue in taxize to explore there https://github.com/ropensci/taxize/issues/628

[zachary-foster]

• and I'd rather not have variable output formats

Yea, variable output formats are not great