Clonal Evolution

Giesa, Spivak, & Buehler (2011) — Reoccurring Patterns in Hierarchical Protein Materials and Music: The Power of Analogies

Properties1
tagsmethods, theoretical, conceptual-synthesis

9 min read

Bibliographic Reference

Giesa, T., Spivak, D. I., & Buehler, M. J. (2011). Reoccurring patterns in hierarchical protein materials and music: The power of analogies. BioNanoScience, 1(3), 153–161. https://doi.org/10.1007/s12668-011-0022-5

Note: This is a short (~9 page) methods paper introducing category theory-based “ontology logs” (ologs) as a rigorous mathematical framework for constructing and validating cross-domain analogies. The paper is a proof-of-concept; the single worked example (spider silk ↔ classical music) demonstrates the formalism rather than surveying “reoccurring patterns” across multiple systems.

Core Argument

Cross-domain analogies — mapping structural patterns from one field to another — are ubiquitous in science but typically informal and unvalidated. The authors propose hierarchical ontology logs (ologs) based on category theory as a rigorous mathematical framework for representing hierarchical structure-function relationships and constructing structure-preserving analogies (functors) between domains. An olog is a category whose objects are sets representing entities in a domain and whose arrows represent unique functions between them. A hierarchical olog adds a subcategory H forming a forest (collection of trees) that imposes hierarchical organization. A functor between two ologs preserves the compositional structure — if paths commute in the source olog, their images must commute in the target olog. This commutativity condition is the formal criterion for a valid analogy.

The paper demonstrates the framework with spider silk protein materials and classical music: amino acids → β-sheet nanocrystals → protein networks → macroscale fiber in silk are mapped functorially to sound waves → motifs → phrases → movements in music. The two ologs are structurally isomorphic under this functor. The authors acknowledge that the framework is descriptive, that graph theory would suffice for structural description, and that category theory’s added value lies in the functorial mapping that validates cross-domain knowledge transfer.

Methods

This is a theoretical methods paper with no experimental component. The methodology consists of:

  • Olog construction: Define objects (sets of entities) and arrows (functions) representing a domain’s structure-function relationships. Build commutative diagrams enforcing uniqueness of paths.
  • Hierarchical ologs: Equip the olog with a subcategory H forming a forest (tree structure), where hierarchy-boxes surround constituents to represent system states at each level. The condition Ob(C) = Ob(H) and H being a forest ensures that hierarchy is imposed via morphism inclusion.
  • Functorial analogy: Define a functor F: Olog_source → Olog_target that maps objects to objects and arrows to arrows, preserving composition. If the functor is an isomorphism, the two domains share identical hierarchical structure. The authors note this is the case for the silk-music analogy: “the positions of boxes and arrows are the same in both systems.”
  • Computational implementation: Ologs are described as “easily implemented in object-based computer languages” via database frameworks (from Spivak & Kent, 2011), but no code, schemas, or queries are provided in this paper.

The linguistic example (phonemes → words → sentences) serves as an intermediate pedagogical illustration bridging the formalism to the biological/musical application.

Key Findings

  • Hierarchical patterns recur across seemingly disparate domains. Spider silk and classical music share a structurally identical hierarchical architecture: basic building blocks (amino acids / sound waves) assemble into intermediate structures (β-sheet nanocrystals / motifs), which compose into higher-level functional units (protein networks / phrases), which form macroscale systems (fiber / movements). The hierarchical olog captures this shared structure as a functorial isomorphism. (Giesa et al., 2011, Sections 3–4)
  • Category theory provides formal criteria for valid analogies. The commutativity condition — if two paths in the source olog point to the same instance, their images in the target olog must also point to the same instance — is the formal criterion for a structure-preserving mapping. This distinguishes valid analogies (where structure is preserved) from mere metaphors. (Giesa et al., 2011, Section 2)
  • Hierarchical interaction is a non-trivial structural pattern. A property of a higher-level structure can relate to an element of a lower-level structure — for example, enzyme catalytic activity (protein-level property) depends on specific amino acid arrangements (building-block-level structure). This pattern, which the paper formalizes through cross-level arrows, captures why driver mutations (lower-level genetic changes) affect clone fitness (higher-level population property). (Giesa et al., 2011, Section 3, Figure 2)
  • The framework explicitly acknowledges its limitations. The authors note that category theory’s added value over graph theory lies in the functorial mapping, not the structural description; that protein folding is non-deterministic and requires specification of environmental conditions; and that artistic properties (musical quality) are inherently subjective and cannot be formally validated by the olog. This candor is methodologically responsible. (Giesa et al., 2011, Sections 3–4)

Concepts Introduced or Used

  • Ontology log (olog) — a category-theoretic knowledge representation where objects are sets of entities and arrows are unique functions between them. Commutative diagrams enforce consistency. Designed for database implementation and cross-domain knowledge transfer.
  • Hierarchical olog — an olog equipped with a subcategory H forming a forest (trees), where hierarchy-boxes surround constituent sets to represent system states at each hierarchical level. The forest structure captures the unique assignment of lower-level elements to higher-level structures.
  • Functorial analogy — a cross-domain analogy formalized as a functor F: Olog_source → Olog_target that preserves compositional structure. The commutativity condition (commuting paths in source must commute in target) is the formal validity criterion.
  • Hierarchical interaction — a structural pattern where a property of a higher-level structure relates to an element of a lower-level structure, crossing the hierarchical boundaries that usually separate composition levels.
  • Materiomics — the holistic study of material systems across scales, integrating structure, function, and properties from nano to macro. The domain for which ologs are proposed as a knowledge-representation tool.

Entities Referenced

  • Markus J. Buehler — Professor of Civil and Environmental Engineering, MIT. Director of the Laboratory for Atomistic and Molecular Mechanics (LAMM). Leading figure in materiomics and the application of category theory to materials science.
  • David I. Spivak — Mathematician at MIT (now at Topos Institute). Developer of the olog formalism (Spivak & Kent, 2011). Primary author of the category-theoretic foundations used in the paper.
  • Category theory — branch of mathematics originating with Eilenberg & Mac Lane (1940s). Studies mathematical structure through objects and morphisms (structure-preserving maps). The paper uses a thin subset: categories, subcategories, functors (in an abused/relaxed sense), and isomorphisms of directed graphs.
  • Spider silk — a hierarchically structured protein material. The paper analyzes it as: amino acid sequences (primary structure) → β-sheet nanocrystals (secondary/tertiary) → protein networks → macroscale fiber. Known for exceptional strength and toughness arising from hierarchical architecture.

Limitations

As stated by the authors and identified by reviewers:

  • The isomorphism is trivial. The silk and music ologs were constructed to be structurally identical — “the positions of boxes and arrows are the same in both systems.” The analogy is built into the representation rather than discovered by the formalism. A genuine test would require independently constructed ologs and a search for non-trivial functorial mappings.
  • Category theory is ornamental. The paper deploys almost none of category theory’s expressive power: no natural transformations, no limits or colimits (mentioned but never used), no adjunctions, no universal properties. “Graph theory would actually provide sufficient means” for the structural description (authors’ own admission). Category theory’s unique value — functorial mappings — is demonstrated only for the trivial isomorphic case.
  • Single proof-of-concept. The paper’s title promises “reoccurring patterns” across multiple domains, but delivers exactly one analogy (silk ↔ music). Table 1 lists further candidates but never formalizes them. Generalizability is not demonstrated.
  • No empirical validation or predictions. The analysis produces no testable hypotheses, no quantitative predictions, and no new material property. It remains a descriptive exercise. For a journal in applied materials science, the absence of empirical grounding is notable.
  • Computational implementation is aspirational. Despite claiming ologs are “easily implemented in object-based computer languages” and “correlate one-to-one to a computer implementation,” no code, database schema, or machine-readable specification is provided.
  • No comparison to existing frameworks. The paper does not compare ologs to established formal ontology frameworks (OWL, description logics, RDF, formal concept analysis). The claim that ologs are “superior” is unsubstantiated.
  • Static representation. Ologs capture knowledge structure but not temporal dynamics. Clonal evolution is inherently temporal (clone frequency dynamics, sweep times). The framework would require extension to time-indexed categories for evolutionary applications.
  • Deterministic assumption. Category theory is silent on probability. Evolution is dominated by stochastic drift. A probabilistic or Markov-categorical extension would be needed for full evolutionary applicability.

Relevance to Clonal Evolution

This paper provides the methodological toolkit for the wiki’s cross-domain synthesis program. The existing concept pages compression-progress-evolution and dual-regime-evolution map concepts across domains (Schmidhuber → evolution, Gabora → cancer) using informal tables and prose. Buehler’s framework provides the formal criterion for when those analogies are valid: a functorial mapping must preserve compositional structure, as verified by commutativity conditions.

Specific contributions to the wiki’s methodology:

  • Hierarchical ologs ↔ cancer genome hierarchy. The paper’s forest-structured hierarchical olog maps directly onto phylogenetic trees (truncal → shared branch → private leaf mutations). The commutativity condition — “tracing a mutation up to the tumor level through two different paths must yield the same result” — is the categorical encoding of phylogenetic consistency. When commutativity fails, it flags convergent evolution, polyclonal origins, or sampling artifact.
  • Functorial validation of the compression-evolution analogy. The functor G: ClonalCat → CompressionCat maps genome states → data, fitness → compression quality, clonal sweeps → compression breakthroughs. The commutativity condition identifies where the analogy is exact (driver mutations under strong selection) and where it breaks (passenger mutations, drift regime, the genome as self-referential compressor).
  • Functorial analysis of the dual-regime model. The functor F: DarwCat → NonDarwCat between genetic and epigenetic levels fails to be a strict functor — epigenetic change is reversible (groupoid), massively parallel, and context-dependent. The failure is informative: the coupling mechanisms identified in dual-regime-evolution (IDH mutation → hypermethylation, MGMT silencing → increased mutation rate) are the coherence conditions that make the non-functorial mapping well-behaved.
  • Discipline for cross-domain work. The paper’s central methodological lesson: an analogy is valid not because it “feels right” but because a functor exists that preserves compositional structure. Every mapping in the wiki’s cross-domain tables should be annotated with its commutativity status.

The paper does not provide concepts for clonal evolution per se — it provides the method for validating the concepts we already have. See compression-progress-evolution § Category-theoretic validation and dual-regime-evolution § Category-theoretic analysis for the integrated synthesis.

Graph View

Backlinks