Agent-Skills.md

Agent Skills: YB Translator

Translate programming concepts to biological parallels using real ontology terms from EBI OLS.

UncategorizedID: plurigrid/asi/yb-translator

Repository

plurigrid/asi
plurigridLicense: MIT
165

Install this agent skill to your local

pnpm dlx add-skill https://github.com/plurigrid/asi/tree/HEAD/plugins/asi/skills/yb-translator

Skill Files

Browse the full folder contents for yb-translator.

Download Skill

Loading file tree…

plugins/asi/skills/yb-translator/SKILL.md

Skill Metadata

Name
yb-translator
Description
Translate programming concepts to biological parallels using real ontology terms from EBI OLS.

YB Translator

Translate programming/CS concepts to biological parallels. Must use real ontology IDs from EBI OLS.

Required Output Format

CONCEPT: [programming concept]
BIOLOGY: [biological parallel]
ONTOLOGY: [Ontology Name] - [Term Name] ([ID])
EXAMPLE: [specific instance from ontology]
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/[ont]/classes/[encoded-iri]

Ontologies to Use

| Ontology | Code | Use For | |----------|------|---------| | Cell Ontology | CL | Cell types, differentiation | | Gene Ontology | GO | Processes, functions, components | | Disease Ontology | MONDO | Disease hierarchies | | Tissue/Anatomy | UBERON | Anatomical structures | | Phenotype | HP | Observable traits | | Pathway | REACT/KEGG | Metabolic/signaling pathways |

Bionty Integration

For programmatic access to biological ontologies, use Bionty:

import bionty as bt

# Lookup GO terms
go = bt.Gene()
go.lookup("RNA polymerase")

# Cell ontology
cl = bt.CellType()
cl.search("T cell")

Bionty provides versioned, validated access to CL, GO, MONDO, UBERON, and more.

Fetch Live Data

bb ~/.claude/skills/yb-translator/scripts/fetch_ontology.clj verify <ID>

Example:

bb ~/.claude/skills/yb-translator/scripts/fetch_ontology.clj verify CL:0000084

Translation Examples

Immutability

CONCEPT: Immutable data structures
BIOLOGY: DNA template strand
ONTOLOGY: Gene Ontology - DNA replication (GO:0006260)
EXAMPLE: Template strand unchanged during replication; new strand synthesized
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0006260

Inheritance/Subtyping

CONCEPT: Class inheritance
BIOLOGY: Cell differentiation hierarchy
ONTOLOGY: Cell Ontology - T cell (CL:0000084)
EXAMPLE: T cell → CD4+ T cell (CL:0000624), CD8+ T cell (CL:0000625)
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/cl/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FCL_0000084

Interface/Protocol

CONCEPT: Interface contract
BIOLOGY: Enzyme classification by function
ONTOLOGY: Gene Ontology - kinase activity (GO:0016301)
EXAMPLE: All kinases transfer phosphate; different substrates
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0016301

Garbage Collection

CONCEPT: Automatic memory management
BIOLOGY: Autophagy
ONTOLOGY: Gene Ontology - autophagy (GO:0006914)
EXAMPLE: Lysosomal degradation of cytoplasmic components
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0006914

Static Type Checking

CONCEPT: Compile-time type verification
BIOLOGY: Receptor-ligand specificity
ONTOLOGY: Gene Ontology - receptor binding (GO:0005102)
EXAMPLE: Insulin receptor (INSR) only binds insulin; shape verified before signal
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0005102

Recursion

CONCEPT: Self-referential function
BIOLOGY: Fractal branching morphogenesis
ONTOLOGY: Gene Ontology - branching morphogenesis (GO:0001763)
EXAMPLE: Lung bronchi: branch → branches → branches (same pattern each level)
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0001763

Concurrency

CONCEPT: Parallel execution
BIOLOGY: Parallel metabolic pathways
ONTOLOGY: Gene Ontology - metabolic process (GO:0008152)
EXAMPLE: Glycolysis and beta-oxidation run simultaneously in cytoplasm/mitochondria
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0008152

Preconditions (Dafny requires)

CONCEPT: Function precondition
BIOLOGY: Enzyme substrate specificity
ONTOLOGY: Gene Ontology - substrate-specific channel activity (GO:0022838)
EXAMPLE: Lactase only accepts lactose; wrong substrate = no reaction
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0022838

Postconditions (Dafny ensures)

CONCEPT: Function postcondition
BIOLOGY: Enzyme product guarantee
ONTOLOGY: Gene Ontology - catalytic activity (GO:0003824)
EXAMPLE: Lactase guarantees galactose + glucose output from lactose input
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0003824

Formal Verification (Dafny)

CONCEPT: Compile-time proof of correctness
BIOLOGY: Immune checkpoint verification
ONTOLOGY: Gene Ontology - T cell activation (GO:0042110)
EXAMPLE: T cell requires MHC presentation + costimulation; verified before response
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0042110

REPL (Clojure)

CONCEPT: Interactive evaluation loop
BIOLOGY: Adaptive immune response
ONTOLOGY: Gene Ontology - adaptive immune response (GO:0002250)
EXAMPLE: Encounter antigen → test response → remember successful patterns
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0002250

Homoiconicity (Clojure)

CONCEPT: Code as data
BIOLOGY: Self-replicating RNA polymerase ribozymes
ONTOLOGY: Gene Ontology - RNA polymerase activity (GO:0097747)
EXAMPLE: Ribozymes catalyze synthesis of copies of themselves—RNA is simultaneously 
         catalyst (program) and template (data). E.g., R3C ligase (Joyce 2002).
SOURCE: https://www.ebi.ac.uk/ols4/ontologies/go/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0097747

Rules

  1. Always include ontology ID (e.g., CL:0000084, GO:0006914)
  2. Always include SOURCE URL to EBI OLS
  3. Use real terms - verify they exist at ebi.ac.uk/ols4
  4. One parallel per concept - no tables, no frills
  5. Run fetch script when uncertain about term existence

<<<<<<< HEAD

Related Skills (Random Walk r=3)

| Skill | Relation | Trit | |-------|----------|------| | assembly-index | Cronin's molecular complexity → code complexity | ⊕ | | alife | Artificial life / origin of life parallels | ○ | | turing-chemputer | Chemical computation ↔ biological computation | ⊖ |

Cross-comparison: All three skills explore the code↔chemistry↔life boundary that yb-translator maps via ontologies.

Related Skills

  • biopython (⊕): Programmatic access to biological data; yb-translator provides the ontological frame, biopython provides the runtime
  • alife (○): Artificial life / origin of life parallels — overlaps on the code↔biology boundary but alife generates, yb-translator classifies
  • assembly-index (⊖): Cronin's molecular complexity metrics; complementary measure to ontological depth

These three are the only skills with direct biological ontology dependencies. Most skills in the repo have no biological structure — that's correct.

SDF Interleaving

This skill connects to Software Design for Flexibility (Hanson & Sussman, 2021):

Primary Chapter: 4. Pattern Matching

Concepts: unification, match, segment variables, pattern

GF(3) Balanced Triad

yb-translator (⊕) + SDF.Ch4 (○) + assembly-index (⊖) = 0

Skill Trit: +1 (PLUS - generation)

yb-translator generates translations (⊕). SDF Ch4 pattern matching provides the structural recognition (○). assembly-index measures complexity of the result (⊖). The triad balances because generating a translation, recognizing its structure, and measuring its complexity are three independent operations that compose to a closed loop.

Secondary Chapters

  • Ch3: Variations on an Arithmetic Theme (combinators for building translators)
  • Ch7: Propagators (bidirectional constraint flow between CS and biology domains)

Connection Pattern

Pattern matching unifies a programming concept with a biological term by structural match on the mechanism — not by keyword. The translator IS a pattern matcher: input a CS concept, output the ontology term whose mechanism matches. When the match is analogical rather than direct, the pattern matcher should say so.

Verification

# Self-verify translations (L0: syntactic + parallel strength)
bb scripts/random_walk_verifier.clj verify

# Inventory all ontology IDs across SKILL.md and examples
bb scripts/random_walk_verifier.clj ids

# Analyze which skills in the repo have actual biological structure
bb scripts/analyze_all_skills.clj

# Check a specific ontology ID against live EBI OLS
bb scripts/fetch_ontology.clj verify CL:0000084

What was rebuilt

The previous analyze_all_skills.clj regex-matched any description into a category theory concept (99% positive rate). The previous random_walk_verifier.clj "verified" those via more regex on the same content. Both were removed.

The new versions:

  • analyze_all_skills.clj — checks for EXPLICIT typed hierarchies or dual ontological contexts. Most skills get :none. If >20% are classified as having structure, the criteria are too loose.
  • random_walk_verifier.clj — verifies yb-translator's OWN translations, not other skills. Checks ontology ID format (L0), flags parallel strength as :direct, :analogical, or :metaphorical. L1 (live EBI check) delegated to fetch_ontology.clj.
  • skill_taxonomy.edn — lists the 12 translations with honest parallel strength ratings. 6 direct, 6 analogical, 0 metaphorical.

origin/main