Use MIxS `samp_taxon_id` to model NCBI taxonomy ID from GOLD?

[sujaypatil96]

Add a slot called ncbi_taxon_id to capture the host taxonomy ID as present in GOLD, and other sources possibly.

For example, let's look at Gb0291745 on the GOLD website: https://gold.jgi.doe.gov/biosample?id=Gb0291745

Look under the Host Metadata section, and you'll see a value of 3689 associated with the host taxonomy ID.

We need a slot under the Biosample class to capture this information. In this issue, I'm proposing the addition of a slot called ncbi_taxon_id generally, and asserting its usage in the Biosample class using it's slots property.

[sujaypatil96]

Comments from discussion with @turbomam:

There is a slot in MIxS called specific_host already, which models the taxonomy name or id.

We have two approaches here:

• make changes to the mixs.yaml file in nmdc-schema repo, add in specific_host slot, import that into Biosample class in nmdc.yaml and constrain specific_host under the slot_usage for Biosample
• add a new slot in nmdc.yaml called ncbi_taxon_id which is inherited from specific_host using the is_a property, constrain the slot and then assert it on the Biosample class

[ssarrafan]

@sujaypatil96 is this something you're currently working on? Can I add it to the sprint board for January?

[sujaypatil96]

Model this similar to ENVO terms. Followup with GSC to figure out why there are three similar terms for taxon id.

[sujaypatil96]

Following up on this, the three MIxS terms with confusing definitions are:

• specific_host: https://genomicsstandardsconsortium.github.io/mixs/0000029/
• host_taxid: https://genomicsstandardsconsortium.github.io/mixs/0000250/
• samp_taxon_id: https://genomicsstandardsconsortium.github.io/mixs/0001320/

[ssarrafan]

@sujaypatil96 I'll move this to the next sprint but let me know if you won't be actively working on it for the next couple of weeks

[sujaypatil96]

@ssarrafan yup, we plan to address this at the metadata call today.

[sujaypatil96]

The information that we intend to capture in the schema from GOLD is the NCBI taxonomy ID (this is the label for the field that appears on the JGI GOLD website)

For example, below is a snippet of the output from the GOLD API:

{'biosampleGoldId': 'Gb0291653', 'biosampleName': 'Bulk soil microbial communities from poplar common garden site in 
Clatskanie, Oregon, USA - BESC-86-CL2_69_17', 'ncbiTaxId': 410658, 'ncbiTaxName': 'soil metagenome', 
'sampleCollectionSite': 'Bulk Soil', 'geographicLocation': 'USA: Oregon'...}

Semantic considerations

• From the JGI GOLD website: https://gold.jgi.doe.gov/biosample?id=Gb0291653 the field label that corresponds to ncbiTaxId is NCBI Taxonomy ID.
• The GOLD page provides a link to a page on the NCBI taxonomy browser page with the following URL: https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=410658
• The above page has the label: soil metagenome

"soil metagenome" is not a host, so neither of the MIxS terms with the word "host" is applicable, leaving only the samp_taxon_id term from the original 3 candidates (specific_host, host_taxid and samp_taxon_id)

Format considerations

Looking at the ncbiTaxId field, it seems to be an integer value. And here is an example value for the samp_taxon_id field that MIxS provides:

Gut Metagenome [NCBI:txid749906]

We could reconstruct a value that looks like the above example using two GOLD fields - ncbiTaxName and ncbiTaxId. We propose replacing the syntax implied by the MIxS example above with syntax like: NCBITaxon:410658 (as found on OLS)

[mslarae13]

• samp_taxon_id

Agree, samp_taxon_id is the right slot. This is NOT collected by the user or in the submission portal, but I think that’s fine. This can just be a GOLD assigned field

• specific_host: https://genomicsstandardsconsortium.github.io/mixs/0000029/
• host_taxid:

These descriptions are more or less forward and reverse, so how they're different is unclear, but I think this is a different issue.

Decision during Wednesday 1pm metadata meeting

Format change ncbiTaxName to NCBITaxon:#### Capture from GOLD “ ncbiTaxName [NCBITaxon:ncbiTaxID] “ example gut metagenome [NCBITaxon:749906] or soil metagenome [NCBITaxon:######]

[turbomam]

Thanks for the notes, @mslarae13

A couple of questions:

1. You're suggesting that there wouldn't be any column for samp_taxon_id in the Submission Portal. How would that relate to our vision of gathering sample metadata in bulk form external sources (like GOLD) and then loading it into the submission portal for a "data wrangler" to check?
2. what does "descriptions are more or less forward and reverse" mean?
3. Could the last two lines in your comment be summarized as below?

---

• NMDC will populate samp_taxon_id slots for Biosamples by composing the ncbiTaxName and ncbiTaxId. For the taxon identifier portion, MIxS implies a syntax of ^NCBI:txid[0-9]+$ like 'NCBI:txid749906' but instead NMDC will follow the syntax used by the OBO foundry, ^NCBITaxon:[0-9]+$ like 'NCBITaxon:749906'

See regexr.com for experimenting with those patterns