microbiomedata / mixs Goto Github PK

This is different from deep diffing the YAML output of various builds. The deep diff CLI doesn't do a merge over LinkML imports

• https://github.com/perma-id/w3id.org/tree/master/gensc
  • "owned" by Ramona
  • this is the real namespace for mixs terms
• https://github.com/perma-id/w3id.org/tree/master/mixs
  • "owned" by @sujaypatil96
• https://github.com/perma-id/w3id.org/tree/master/nmdc
  • "owned" by @turbomam

At least some of these if statements probably need to be de-indented

Ensure that the slot usage can be used to reconstruct the -, C, E, M or X value

Bioscales has a column for nitrogen, expressed as a percent

Here are all the MIxS terms containing 'nit' in their term names aka Structured comment names

• normalize tokens
• document delimiters and grammar

See static slot analysis

Could start by making them annotations

Analytes, etc.

• Ammonium
• Carbon
• Nitrate
• Nitrite
• Nitrogen
• Phosphate
• Phosphorus
• Temperature

Modifiers

• air
• biomass
• dissolved
• inorganic
• method
• microbial
• organic
• particulate
• reactive
• soluble
• total

Additional patterns from @mslarae13 ?

Ignore list ?

• antimicrobial
• dioxide
• hydrocarbon
• ratio

For example, in the ...packages... sheet, where terms can be combined with different environments?

Proposal:

• improve definitions for consistency and readability
• scrub garbage characters

including w3id reservations

@sujaypatil96 improved the excel generator recently, so let's try it on MIxS

https://github.com/GenomicsStandardsConsortium/mixs has the repo website set to https://genomicsstandardsconsortium.github.io/mixs/

We might want to change that while we work out w3id redirection, documentation styles, etc.

@sujaypatil96 I'll need help

see

• #51

Bioscales has a carbon column for their soil samples, expressed in percent

Here are the existing MIxS terms containing 'car'

Note agriculture's 'tot_car' vs microbial mat/biofilm's & sediment's 'tot_carb'

Comments from @cmungall:

• full name not just short name
• don’t bother including things like Usage
• leave off mappings for now - there are no mappings in MIxS (yet)
• put descriptions of classes (combos and packages and checklists) on the front page and just include a bit more text to let people know what is going on
• for classes, let’s generate a title that is not camelcase and use that (this is an upstream change though)
  maybe show combos as a grid?
  • eg https://microbiomedata.github.io/mixs/fao_class/#fao_class
• hyperlink URLs
• Maybe “Inheritance” is not so useful - we know everything has max one ancestor here?
• it’s not so useful to end up here - maybe exclude abstract classes?

agriculture's lib_layout is the only term on the "MIxS6 packages - Final_clean" sheet with a section assignment ('sequencing')

• case... are upper case characters ever allowed?
• whitespace, underscores or hyphens?
• slight name variants
• systematic token variants, like meth vs method
• check for flip-flops

MIxS6 packages - Final_clean: https://docs.google.com/spreadsheets/d/1QDeeUcDqXes69Y2RjU2aWgOpCVWo5OVsBX9MKmMqi_o/edit#gid=750683809

see also #10

that really applies to all of our LinkML repos

the mixs-source output says

  barometric_press:
    is_a: core field 

But barometric_press is in the MIxS6 packages - Final_clean, not the MIxS6 Core- Final_clean sheet

https://docs.google.com/spreadsheets/d/1-ocpwjx6nkBod6aj4kcYeSB5NRlhXaYCcuk3ooX2OV4

Early on, I converted MIxS Value syntaxes to LinkML string_serializations. @cmungall also parses the Value syntaxes to detect potential enumerations. There are some hybrid Value syntaxes that get converted into enums with mangled permissible values that are especially problematic when serializing to RDF.

string_serializations have two possible applications:

1. combining the contents of two separate fields into one, based on the pattern. I believe @sujaypatil96 has implemented that in linkml-convert.
2. parsing the contents of a field. I don't believe that has been implemented anywhere, although it would be handy for making the conversion of DataHarmonizer output into schema-compliant JSON more declarative.

structured_patterns can be used to assemble complex regular expressions from reusable components, like the {}-wrapped tokens from MIxS Value syntaxes. The complex regular expressions can then be used to validate input into DataHarmonizer.

I think I should switch the instantiation of MIxS Value syntaxes from string_serializations to structured_patterns

see #42

CC @sujaypatil96

Bioscales has a P column, with values 111.4 mg/kg (ppm)

@mslarae13 has raised some concerns about errors or ambiguity in these slots:

note term MIXS:0000689 tot_phos from agriculture

• all other packages have a tot_phosp term
• the Package Item 'total phosphorous' and the Definition `'Total amount or concentration of phosphate' are incompatible

nobody should ever want to look at the spreadsheets

searches like pH, class and lat (from lat_lon) should bring most relevant terms to the top. we're probably more interested in term titel and description.

• Environmental package
• Structured comment name
• Package item
• Definition
• Expected value
• Value syntax
• Example
• Section
• Requirement
• Preferred unit
• Occurrence
• MIXS ID
• github ticket

Problems with cardinality

• Several mixs_notations error out
• Several cardinality maps_tos don't do anything… conditional not implemented?
  • can't round-trip cardinality?

north, east, south, west, etc.

What value was added

• checklist mixins (with some annotations?)
• MIxS sections modeled as LinkML parent slots
• are subsets being used (check l2s supplement from linkml-abuse)

Any surprises?

• some slots have "core" as a parent, even though they aren't defined in the core sheet

MIxS 6 term updates?

• MIxS6 Core- Final_clean: https://docs.google.com/spreadsheets/d/1QDeeUcDqXes69Y2RjU2aWgOpCVWo5OVsBX9MKmMqi_o/edit#gid=178015749
• MIxS6 packages - Final_clean: https://docs.google.com/spreadsheets/d/1QDeeUcDqXes69Y2RjU2aWgOpCVWo5OVsBX9MKmMqi_o/edit#gid=750683809

Minimum Information about any (X) Sequence” (MIxS) specification