Skip to content

AGENTS.md

Latest commit

 

History

History
89 lines (78 loc) · 2.93 KB

AGENTS.md

File metadata and controls

89 lines (78 loc) · 2.93 KB
Raw

MOKG Configuration

Generated: 2026-03-01 Commit: c8bb97a Purpose: Tablassert TC3 table configurations for MicrobiomeKG and MultiomicsKG knowledge graphs

OVERVIEW

YAML-based table configurations using TC3 format for extracting knowledge graph assertions from research publications. Configs are processed by Tablassert and deployed to NCATS Translator ecosystem via PloverDB.

STRUCTURE

./
├── GRAPH/          # Graph-level orchestration configs
├── TABLE/          # Individual table configurations
│   ├── MBKG/        # MicrobiomeKG tables (53 files)
│   ├── MIGRATASSERT/  # Migration-related assertions (49 files)
│   ├── MOKG/         # MultiomicsKG tables (21 files)
│   └── V6/           # V6 build tables (7 files)
└── README.md

WHERE TO LOOK

Task Location
MicrobiomeKG tables TABLE/MBKG/
Migration assertions TABLE/MIGRATASSERT/
MultiomicsKG tables TABLE/MOKG/
Graph orchestration GRAPH/V6/MOKGV6.yaml

TC3 CONFIG FORMAT

Required fields:

template:
  syntax: TC3
  source:
    kind: excel|text
    url: <download URL>
    local: <cached file path>
    sheet: <sheet name>         # excel only
    row_slice: [start, auto]       # optional
  statement:
    subject:
      method: column|value
      encoding: <column name or literal>
      prioritize: [Biolink categories]
      remove: [regex patterns]
      regex: [{pattern, replacement}]
    predicate: <Biolink predicate>
    object:
      method: column|value
      encoding: <column name or literal>
      prioritize: [Biolink categories]
  provenance:     # NOTE: spelled "provenance" in this project
    repo: PMC|PUBMED|DOI
    publication: <identifier>
    contributors:
      - kind: curation|validation|tool
        name: <curator name>
        date: <date>
  annotations:
    - annotation: <attribute name>
      method: column|value
      encoding: <column name or literal>

NodeEncoding options:

  • taxon: <int> - NCBI Taxon ID for filtering (9606=human, 10090=mouse)
  • explode_by: <delimiter> - Split delimited values into separate edges
  • fill: forward|backward|min|max|mean|zero|one - Null handling strategy

CONVENTIONS

  • Source URLs: Point to PMC/DOI for published data
  • Local paths: Relative to ./DATALAKE/ (not present in repo)
  • Provenance spelling: Uses provenance: (not provenance)
  • Curation metadata: Always include contributor with name + date + organization

ANTI-PATTERNS

  • Missing provenance: section → invalid configuration
  • Using provenance: (standard spelling) instead of provenance:

COMMANDS

No build/test commands - YAML files consumed directly by Tablassert.

NOTES

  • Configs use PMC article IDs for publication references
  • Taxon filtering available via taxon: field
  • Regex transformations apply in order
  • Annotation examples: p-value, sample size, correction method, relationship strength