Run any Skill in Manus with one click

map-media-ingredients

Stars1

Forks0

UpdatedJune 21, 2026 at 02:12

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.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

CultureBotAI

CultureBotAI/MediaIngredientMech

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Biological Scientists, All OtherLife, Physical, and Social Science Occupations·SOC 19-1029

File Explorer

5 files

SKILL.md

readonly

More from this repository

same repository

manage-identifiers

CultureBotAI/MediaIngredientMech

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.)

2026-06-241

merge-ingredients

CultureBotAI/MediaIngredientMech

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.

2026-06-241

review-ingredients

CultureBotAI/MediaIngredientMech

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.

2026-06-201

id-label-correspondence

CultureBotAI/MediaIngredientMech

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.

2026-06-121

deep-research-ingredient

CultureBotAI/MediaIngredientMech

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.

2026-05-291

schema-gap-analysis

CultureBotAI/MediaIngredientMech

Find gaps between MIM's LinkML schema, its YAML instances, and the code that generates them. Uses linkml-validate (or the strict closed-schema harness ported from CultureMech) as ground truth and reports along three axes (schema / instances / process). Copy-paste runnable.

2026-05-241

name	map-media-ingredients
description	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.
category	workflow
requires_database	false
requires_internet	true
version	1.1.0

Media Ingredient Ontology Mapping

Overview

Map microbial growth-media ingredient names to authoritative ontology terms for semantic integration and knowledge-graph construction. This ensures consistency across datasets, enables semantic queries, and links to KG-Microbe / CHEBI / FOODON / ENVO. Scope: chemical compounds (salts, organics), biological materials (extracts, peptones), and environmental samples (soil, seawater).

When to Use This Skill

Adding new ingredients to MediaIngredientMech, or curating unmapped ones from formulations
Validating existing ontology mappings
Integrating media data with the KG-Microbe knowledge graph
Normalizing chemical names with hydrates, catalog codes, or incomplete formulas
Deciding which ontology (CHEBI vs FOODON vs ENVO) to use

Ontology Selection Guide

Ingredient Type	Primary Ontology	Examples	When to Use
Simple chemicals	CHEBI	NaCl, glucose, MgSO4•7H2O, K2HPO4	Pure chemical compounds, salts, ions
Biological materials	FOODON	yeast extract, peptone, tryptone, beef extract	Biological/food-derived preparations
Environmental samples	ENVO	soil extract, seawater, sediment	Natural environmental materials
Complex mixtures	often unmappable	"Vitamin solution A", "Trace metals"	Too generic for a specific mapping

Priority order: CHEBI → FOODON → ENVO. Try the most specific ontology first; fall back to broader ones.

Chemical Normalization Rules

Many names need normalization before matching — chemical_normalizer.py handles these patterns, and preserves the original form as a typed synonym:

Hydrate stripping (synonym HYDRATE_FORM): MgSO4•7H2O → MgSO4 → "magnesium sulfate"; CaCl2·2H2O → "calcium chloride".
Incomplete-formula correction (synonym INCOMPLETE_FORMULA): K2HPO → K2HPO4 → "dipotassium phosphate".
Catalog-number removal (synonym CATALOG_VARIANT): NaCl (Fisher S271-500) → NaCl.
Abbreviation expansion: dH2O → "distilled water"; NaOAc → "sodium acetate".
Formula → common name: built-in FORMULA_TO_NAME map (40+ entries, e.g. NaCl → "sodium chloride", MgSO4 → "magnesium sulfate").

Matching Strategies

Escalate from cheapest to most expensive; use the simplest that works. Full code (OAK, OntologyClient, chemical_normalizer, OLS) is in reference/matching-strategies.md.

Level 1 — Exact match. Case-insensitive string match against ontology labels, via OAK basic_search or OntologyClient.search(..., sources=["CHEBI"]).
Level 2 — Normalized match. Apply normalize_chemical_name / generate_search_variants, then client.search_with_variants(...) (deduplicates, keeps best scores).
Level 3 — Fuzzy match. OAK lexical search (l~ prefix) or the EBI OLS web API for synonyms/spelling variants. OAK is primary (fast, offline); OLS is secondary (cross-validation).
Level 4 — Manual curation. For complex mixtures, generic/incomplete terms, ambiguous names, or novel formulations — mark NEEDS_EXPERT.

Workflows

A. Interactive single ingredient

python scripts/curate_unmapped.py

Normalizes the name, generates variants, searches CHEBI → FOODON → ENVO, presents scored candidates with media-usage context; user accepts or marks NEEDS_EXPERT.

B. Batch curation (auto-curable simple chemicals)

python scripts/analyze_unmapped.py                                    # categorize first
python scripts/batch_curate_unmapped.py --category SIMPLE_CHEMICAL \
  --auto-normalize --min-confidence 0.9 --dry-run                     # preview
python scripts/batch_curate_unmapped.py --category SIMPLE_CHEMICAL \
  --auto-normalize --min-confidence 0.9                               # apply

Best for SIMPLE_CHEMICAL with clear normalization patterns and score ≥ 0.9. Auto-normalizes with synonym preservation, multi-variant dedup, dry-run, and provenance tracking.

C. Claude Code-assisted curation (complex cases, zero API cost)

python scripts/prepare_for_claude_curation.py --category UNKNOWN --limit 20 --output notes/batch_001.md
# Open notes/batch_001.md and ask Claude Code to suggest ontology mappings →
python scripts/apply_claude_suggestions.py --suggestions notes/batch_001_suggestions.yaml --validate

Full reasoning/audit trail for ambiguous cases. See docs/CLAUDE_CODE_CURATION.md.

Quality Guidelines

Quality Level	Meaning
`EXACT_MATCH`	String matches the ontology label exactly
`SYNONYM_MATCH`	Matches a known ontology synonym
`CLOSE_MATCH`	Semantically equivalent, different phrasing
`MANUAL_CURATION`	Expert judgment required
`LLM_ASSISTED`	Claude Code / LLM suggested
`NEEDS_EXPERT`	Requires domain-expert review

All mappings record curator (automated/manual/LLM), timestamp, and reasoning.

Categorization System

scripts/analyze_unmapped.py assigns categories + mappability scores to prioritize effort:

Category	Description	Mappability	Strategy
`SIMPLE_CHEMICAL`	Salts, common organics	High (80–100)	Batch curation
`COMPLEX_MIXTURE`	Vitamin/metal solutions	Low (10–30)	Often unmappable
`ENVIRONMENTAL`	Soil, seawater	Medium (50–70)	Try ENVO
`INCOMPLETE`	Generic terms	Low (20–40)	Manual review
`PLACEHOLDER`	"See source"	Very low (0–10)	Leave unmapped
`UNKNOWN`	Uncategorized	Medium (40–60)	Claude Code-assisted

Best Practices

DO

Normalize before searching; try multiple ontologies (CHEBI → FOODON → ENVO).
Use the most specific term ("D-glucose" > "glucose" > "sugar").
Record provenance and preserve original forms (hydrates, catalogs) as synonyms.
Validate that the term exists and the definition matches; check KG-Microbe first to reuse existing entities.

DON'T

Don't force mappings for complex mixtures, or use generic terms when a specific one exists.
Don't skip normalization for chemicals, or ignore synonym preservation.
Don't guess at low confidence — mark NEEDS_EXPERT; and don't create duplicates.

Reference Files

File	Contents
`reference/matching-strategies.md`	Full code for all 4 matching levels — OAK `basic_search`, `OntologyClient.search`/`search_with_variants`, `chemical_normalizer`, OAK `l~` fuzzy search, and the EBI OLS API wrapper
`reference/kg-microbe-integration.md`	CultureMech import/export scripts, checking existing KG-Microbe ingredients, and how ingredients link to media/organisms
`reference/examples-and-troubleshooting.md`	Five worked examples (hydrate, incomplete formula, biological extract, environmental, unmappable mixture) and troubleshooting (no/too-many/ambiguous matches, complex mixtures)
`reference/tools-and-api.md`	Curation scripts, core utilities, docs, related skills, and the `chemical_normalizer`/`OntologyClient`/`UnmappedCurator` implementation API