菜单
Build a single unified ingredient mapping TSV across CultureMech, MediaIngredientMech, and CommunityMech — one row per unique ingredient name with all available identifiers
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use this skill to manage MediaIngredientMech record identifiers. In MIM the record `identifier` IS the ontology mapping CURIE (CHEBI / FOODON / cas / kgmicrobe.compound / kgmicrobe.ingredient / mesh / NCIT / MICRO / ENVO) for a mapped ingredient, or an `UNMAPPED_NNNN` placeholder for an unmapped one — there is no separate sequential id. Use when adding or importing ingredient records, minting the next `UNMAPPED_NNNN` placeholder, promoting an unmapped record to its resolved CURIE, or reconciling identifier collisions. (Sequential `RepoName:NNNNNN` ids are a sister-repo scheme — see the cross-repo reference.)
Use this skill to deduplicate and merge MediaIngredientMech ingredient records — when two or more records describe the same ingredient (duplicate CHEBI mappings, name variants, or a placeholder plus its resolved record) and must be consolidated into one, preserving synonyms, occurrence statistics, and curation_history. Covers detection strategies, the merge procedure, and post-merge validation.
Use this skill to map MediaIngredientMech growth-media ingredient names to ontology terms (CHEBI for chemicals, FOODON for biologicals, ENVO for environmental) via exact → normalized → fuzzy → manual matching, with chemical normalization (hydrates, formulas, catalog codes) and synonym preservation. Use when curating unmapped ingredients, adding new ones, or validating existing mappings.
Use this skill to quality-check ontology-mapped MediaIngredientMech ingredient records — verify CHEBI/FOODON/ENVO terms exist and labels match (OAK), enrich chemical properties and synonyms (OLS/OWL), and catch dual-identifier, purity-merge, and kg-microbe-dictionary disagreements. Use after a curation batch, before a KG export, or for periodic maintenance. Issues are graded P1 (blocking) → P4 (optional); P3/P4 can be auto-corrected.
Validate that every ontology ID carries its correct ontology LABEL across MIM data inputs, intermediates, and data products (SSSOM/CSV). Uses LinkML schema bindings + linkml-term-validator for YAML data (canonical label) and a shared OAK validator for products (canonical-or-synonym). Run report-then-enforce; triage drift as wrong-label vs wrong-id.
Run Edison Scientific deep research (PaperQA3) for a MediaIngredientMech ingredient record to gather source-backed identity, composition, CAS-RN, and ontology-grounding (CHEBI/FOODON/NCIT) evidence. Captures a full provenance bundle and produces a curation-focused report for a curator to review and apply.
| name | build-unified-mapping |
| description | Build a single unified ingredient mapping TSV across CultureMech, MediaIngredientMech, and CommunityMech — one row per unique ingredient name with all available identifiers |
| category | integration |
| requires_database | false |
| requires_internet | false |
| version | 1.0.0 |
| tags | ["unified-mapping","cross-repo","integration","tsv","ingredients","chebi","cas-rn","kg-microbe"] |
Purpose: Produce a single canonical TSV (and YAML summary) that maps every unique ingredient name observed across CultureMech media to all available identifiers from MediaIngredientMech.
Output columns:
| Column | Description |
|---|---|
ingredient_name | Name as it appears in CultureMech media |
culturemech_term_id | Ontology term ID recorded in CultureMech (term.id; CHEBI/FOODON) |
occurrence_count | How many CultureMech media use this ingredient |
mim_id | MediaIngredientMech record identifier (CHEBI:XXXXX or UNMAPPED_XXXX) |
chebi_id | ChEBI ontology ID |
cas_rn | CAS Registry Number |
kg_microbe_node_id | KG-Microbe CHEBI node ID (if ingredient present in KG graph) |
mapping_status | MAPPED / UNMAPPED / HAS_TERM_ID_NO_MIM / UNMATCHED_IN_MIM |
example_media | Up to 3 CultureMech media IDs using this ingredient |
Match logic (priority order):
term.id (CHEBI) → look up in MIM by CHEBI ID# From culturebotai-claw repo (recommended — it has the script)
cd ~/Documents/VIMSS/ontology/KG-Hub/KG-Microbe/culturebotai-claw
# Default: both TSV and YAML summary in workspace/
python scripts/build_unified_ingredient_mapping.py
# Custom output path
python scripts/build_unified_ingredient_mapping.py --output workspace/my_mapping.tsv
# TSV only
python scripts/build_unified_ingredient_mapping.py --format tsv
# Or via justfile
just build-unified-mapping
Output files:
workspace/unified_ingredient_mapping.tsv — full flat table (one row per unique CultureMech ingredient name)workspace/unified_ingredient_mapping.yaml — summary with status breakdown and top 50 fully-mapped ingredients| Metric | Count | % of 3,853 unique names |
|---|---|---|
| Matched to MIM record | 1,185 | 30% |
| Have CHEBI ID | 1,212 | 31% |
| Have CAS-RN | 1,107 | 28% |
| Have KG-Microbe node ID | 775 | 20% |
| Fully mapped (CHEBI + CAS) | 1,094 | 28% |
The ~70% not matched to MIM are largely: complex trace mineral stocks, undefined buffer solutions, duplicate name variants (hydrate forms), catalog-coded entries, and FEBA-specific formulations that haven't been curated into MIM yet.
mapping_status| Status | Meaning |
|---|---|
MAPPED | In MIM with full ontology mapping |
UNMAPPED | In MIM but lacking ontology term (unmapped record) |
HAS_TERM_ID_NO_MIM | CultureMech has a term.id but no MIM record |
UNMATCHED_IN_MIM | Not found in MIM by name or CHEBI |
To increase the match rate:
just feba-enrich-ontology in culturebotai-clawjust feba-create-mim-ingredients in culturebotai-clawpython scripts/enrich_mim_cas_rn.py in culturebotai-claw1. Run the script
python scripts/build_unified_ingredient_mapping.py
2. Review coverage
# Count fully-mapped rows
awk -F'\t' '$5!="" && $6!=""' workspace/unified_ingredient_mapping.tsv | wc -l
# Find unmapped high-frequency ingredients
awk -F'\t' 'NR>1 && $9=="UNMATCHED_IN_MIM" {print $3"\t"$1}' \
workspace/unified_ingredient_mapping.tsv | sort -rn | head -20
3. Commit the output (optional — it's in workspace/ which is gitignored by default)
cp workspace/unified_ingredient_mapping.tsv UNIFIED_INGREDIENT_MAPPING.tsv
git add UNIFIED_INGREDIENT_MAPPING.tsv
git commit -m "Update unified ingredient mapping"
../culturebotai-claw/scripts/build_unified_ingredient_mapping.pyjust build-unified-mapping in culturebotai-claw../culturebotai-claw/scripts/enrich_mim_cas_rn.py../culturebotai-claw/scripts/match_mim_to_kg.py