Changes in @only1chunts' new XLSX representation of MIxS

[turbomam]

• Note that these revert away from NCBI's structured comment names in several places, presumably to keep the SCN lengths under 20 characters.
• See https://www.ncbi.nlm.nih.gov/biosample/docs/attributes/
• Where does it say that the SCN's length must be less than N characters? (Ideally the source would be a computable file, not a narrative.)
• Also: shouldn't we be consistent in the tokens composed together? See the keyword notes in XXX

see also

• 50

• 51

• 53

---

• GSC 6.1: air_particulate_matter_concentration GSC 6.1: MIXS:0000108 CH: air_PM_concen. NCBI: air_pm_concen
• GSC 6.1: host_family_relation GSC 6.1: MIXS:0000872 CH: host_fam_rel NCBI: host_family_relationship (underspecified)
• GSC 6.1: host_family_relationship GSC 6.1: MIXS:0000872 CH: host_fam_rel NCBI: host_family_relationship (underspecified)
• GSC 6.1: host_infra_specific_name GSC 6.1 MIXS:0000253 CH: host_infra_spec_name NCBI: host_infra_specific_name (24 chars)
• GSC 6.1: host_infra_specific_rank GSC 6.1 MIXS:0000254 CH: host_infra_spec_rank NCBI: host_infra_specific_rank ()
• GSC 6.1: microbial_biomass_meth GSC 6.1: MIXS:0000339 CH: micro_biomass_meth NCBI: host_infra_specific_rank ()
• GSC 6.1: non_mineral_nutr_regm MIXS:0000571 GSC 6.1: non_min_nutr_regm NCBI: non_mineral_nutr_regm ()
• GSC 6.1: previous_land_use_meth GSC 6.1: MIXS:0000316 CH: prev_land_use_meth NCBI: previous_land_use_meth ()
• GSC 6.1: samp_collec_device GSC 6.1: MIXS:0000002 CH: samp_collect_device NCBI: samp_collect_device
• GSC 6.1: samp_collec_method GSC 6.1: MIXS:0001225 CH:samp_collect_method NCBI: samp_collect_method
• GSC 6.1: single_cell_lysis_appr GSC 6.1: MIXS:0000076 CH: sc_lysis_approach NCBI: ???
• GSC 6.1: single_cell_lysis_prot GSC 6.1: MIXS:0000054 CH: sc_lysis_method NCBI: ???

Incomplete salinity remodeling?

• GSC 6.1: salinity_meth GSC 6.1: MIXS:0000341 GSC 6.1: salinity_meth Reference or method used in determining salinity
• GSC 6.1: samp_salinity GSC 6.1: MIXS:0000109

salinity salinity The total concentration of all dissolved salts in a liquid or solid sample. While salinity can be measured by a complete chemical analysis, this method is difficult and time consuming. More often, it is instead derived from the conductivity measurement. This is known as practical salinity. These derivations compare the specific conductance of the sample to a salinity standard such as seawater. measurement value {float} {unit} 25 practical salinity unit X practical salinity unit, percentage 1 MIXS:0000183

[only1chunts]

I'm not entirely sure what all of this ticket means, but one part of it was about seeing the strucutred comment name length written down somewhere, I can answer that bit: Its from the INSDC feature table documentation here https://www.insdc.org/submitting-standards/feature-table/#3.1

3.1 Naming conventions Feature table components, including feature keys, qualifiers, accession numbers, database name abbreviations, and location operators, are all named following the same conventions. Component names may be no more than 20 characters long (Feature keys 15, Feature qualifiers 20) and must contain at least one letter. The following characters are permitted to occur in feature table component names:

• Uppercase letters (A-Z)
• Lowercase letters (a-z) Numbers (0-9)
• Underscore (_)
• Hyphen (-)
• Single quotation mark or apostrophe (')
• Asterisk (*)

It does appear that the BioSamples database are not following the INSDC rules, so perhaps we dont need to either, but thats for a wider discussion as it would be a change in a fundamental policy.

[turbomam]

Thanks, that's clear and helpful.

For compatibility with LinkML, slot/term names should consist only of lowercase letters, numbers and underscores. One could call that lower snake case. Numbers should not appear first.

Using some of those additional characters will break the very most basic LinkML functionality.

Others will make it difficult to generate or understand derived artifacts, like jsonschema and RDF. This is because LinkML casts element names into the legal namespace for each format.

The terms that we have discussed in this matter the most are 16s_recover and 16s_recover_software. They do seem to work with jsonschema. I will try RDF next.

Note that, as of last month, there were no values populated into either the 16s_recover or 16s_recover_software filed in any NCBI Biosample

[turbomam]

I would like to review whether there are slots whose names must contain uppercase letters, like HACCP_term and IFSAC_category. If there's a record that data has already been entered into NCBI with those terms in those cases, I guess that would argue for leaving them as-is.

NCBI's attributes file already casts to lower case:

• haccp_term (no values populated in NCBI BiosampleSet as of last month)
• ifsac_category (125 unique values populated into ~ 2000 Biosamples)

[turbomam]

The names for classes (checklists, environmental packages, extensions) should be PascalCase aka UpperCamelCase. I have just forced all of those already but we should review the consequences of that. The title can always match something historical and non-PascalCase.

[only1chunts]

as long as the slot label (or whatever we call the long version of the name) can handle the correct case text then I dont think the short names matter so much. Any validation tool we end up making should be capable of understanding differences in case.

[turbomam]

Thanks for the immediate feedback, @only1chunts. Aren't you on holiday?

[only1chunts]

I have tomorrow off, and I'm now about to log-off!