claMLST import relies on deprecated pubMLST API DB list

[meg256]

I tested claMLST import and noticed that it does not seem to list all possible pubMLST schemes (for example, E. coli Achtman is missing).

It seems that your get_mlst_species function pulls its scheme list from the pubMLST REST API landing page.

Unfortunately, I suspect that this landing page has not been updated since 2017, when the REST API was first created.

Can I recommend instead populating your scheme list from the pubMLST database XML? Or the static link.

For example, "Escherichia coli#1" aka E. coli Achtman - https://rest.pubmlst.org/db/pubmlst_ecoli_achtman_seqdef

{"loci":"https://rest.pubmlst.org/db/pubmlst_ecoli_achtman_seqdef/loci","curators":"https://rest.pubmlst.org/db/pubmlst_ecoli_achtman_seqdef/curators","schemes":"https://rest.pubmlst.org/db/pubmlst_ecoli_achtman_seqdef/schemes","sequences":"https://rest.pubmlst.org/db/pubmlst_ecoli_achtman_seqdef/sequences"}

[bvalot]

Hi,

In fact I use the rest API and not the xml list. Norammly, this API is up to date and you can for example download E. coli Achtman using the following code. I think it doesn't work for you because it is referenced as escherichia spp. and not escherichia coli
claMLST import -m achtman /tmp/test.db escherichia

I see the xml list, but the site indicated that this list is not maintained and that we must use instead the API:
https://pubmlst.org/data/

This page contains download links to alleles and profiles that define classic MLST schemes. Where these are hosted on PubMLST, they represent a small subset of the data hosted, all of which is available via the application programming interface (API). A few of the schemes are hosted at Pasteur and the links point to their API. This list is not being maintained for new schemes - all data are available via the API and this is the recommended way to retrieve data programmatically.

[meg256]

There is no problem with using the API to get files! That's not the issue. But the API's landing page is deprecated: it contains an incomplete list of schemes, and lists some schemes that are now deprecated.

Here is the list of current pubMLST databases, the same list as the XML: https://pubmlst.org/data

The database list includes two E. coli schemes (#1 Achtman and #2 Pasteur) that don't show up on your API landing page. This is NOT the same as the outdated generic "Escherichia spp." on the API landing page, which you can see by looking at the API db link for each scheme:

Escherichia coli#1: https://rest.pubmlst.org/db/pubmlst_ecoli_achtman_seqdef/schemes
Escherichia coli#2: https://bigsdb.pasteur.fr/api/db/pubmlst_ecoli_seqdef/schemes

Escherichia spp.: https://rest.pubmlst.org/db/pubmlst_escherichia_seqdef/schemes

There are other outdated schemes on the API landing page, such as the outdated generic "Salmonella spp." scheme, which is now replaced by the Salmonella enterica scheme.

[bvalot]

Why the site pubmlst indicated exactly the contrary?
That the page data is not maintained and you must use the API?

[bvalot]

If you look at the link to the database on API from Escherichia spp., you can see atchman profiles:
https://rest.pubmlst.org/db/pubmlst_escherichia_seqdef/schemes/1

And the last updated is 2024-06-10

[bvalot]

If you look on profiles obtained by the 2 ways, you get the same data.

[bvalot]

To be honest, I want to switch to xml files for the listing of profiles and to have the element on the link, but the message on top of the page that indicated it is not maintained make me stay on the old API

[meg256]

But the XML "Escherichia coli#1" links to "pubmlst_ecoli_achtman_seqdef" database/scheme 4, which is the "real" Achtman scheme - currently the most commonly used E. coli mlst scheme.

And the API "Escherichia spp." links to "pubmlst_escherichia_seqdef" database/scheme 1, which is an outdated generic scheme. I do see that this scheme is still updated; but to my knowledge it is no longer widely used and it is not the "real" Achtman E. coli scheme.

The reason I say confidently that these are not the same database/scheme is because I used claMLST import to download "Escherichia spp."/pubmlst_escherichia_seqdef and "Salmonella spp."/pubmlst_salmonella_isolates and did not get correct results when typing my benchmarking set.

Anyway, it is not a big deal because of course I can just manually initialize a database from the profiles/loci fasta files.

Maybe there is some other way to get a full list of the available databases from the API? For what it's worth, Torsten Seeman's mlst tool updates databases by pulling from the xml, and Sanger ariba does so as well.

[bvalot]

I make download of the 2 profiles and it is the same. Using import, do you specify achtman mlst profile?

Anyway, I would consider to use XML file that clearly simplify the bioinformatic process for future release and limit some strange report of false MLST database.