NeXO ontology construction
Algorithms AIMech:NeXOOntologyConstruction DRAFT
Aliases: NeXO construction algorithm
A network-based ontology-construction procedure that starts from a probabilistic hierarchy of gene communities, removes unsupported binary restrictions to permit multiway joins, and adds supported second-parent connections to produce an overlapping term hierarchy.
Identity Notes & Concept Scope
Compared all 442 pinned AIO and 666 native IDs, names, aliases and scoped synonyms for the three candidates and their publication identifiers; no normalized identity match. Ignored- and hidden-inclusive fixed-string searches of the recovery, older discovery and original Documents checkouts returned 26, 0, 0 lines. All matches were read and are prior source/discovery context, with no existing maintained identity. Semantic assessment covered 44 AIO and 260 native definitions: 41 AIO and 228 native rows were unchanged from earlier reading; three additional AIO and 32 additional native definitions were read. Fourteen complete annotated AIO node/edge/generated contexts and the candidate-specific native records were inspected. Retain the 2013 NeXO construction method separately from its output yeast ontology, the upstream HAC-ML guide, CliXO and the associated alignment operation. Its defining additions are internal-node removal with a density guard and heuristic second-parent edges; it does not infer all relation meanings.
Claims & Grounded Evidence
6 Claims
Mechanism
#probabilistic-backbone
The procedure first builds a binary community tree from interaction-network data and associates internal groups with candidate ontology terms. The paper displays a product of Beta-integrated edge/hole factors while calling the optimization maximum likelihood and referring to HAC-ML. HAC-ML’s original paper distinguishes a profiled-likelihood guide from integrated Bayesian model-selection scores. The displayed NeXO objective and the uninspected software optimization should therefore not be silently equated.
Mechanism
#multiway-joins
The method extends earlier top- and bottom-level pruning to internal tree nodes. It compares a parent’s pooled interactions with its siblings against separate interactions between its children and those siblings, using an integrated-probability ratio. Removal additionally requires that interaction density among the children does not exceed that between the parent and its siblings. Replacing the removed node by its children permits a multiway join.
Mechanism
#additional-parents
Starting near the leaves, the heuristic considers a smaller gene set as a child of a larger candidate parent. It excludes pairs already on a common path or sharing a parent, requires sufficient cross-set interaction density and a hypergeometric threshold, and requires the union to retain at least half the parent’s density. Accepted edges permit multiple parents. This is a local heuristic, not an optimization certificate for every possible ontology DAG.
Scope
#ontology-and-label-scope
The resulting term graph and descendant-gene assignments are separate from the particular yeast NeXO ontology produced in the study. Alignment to GO and other evidence can supply names for some terms; unmatched terms can retain systematic identifiers. The paper explores relation-type suggestions but does not establish automatic is-a or part-of assignment for every inferred relation.
Limitation
#robustness-formula-boundary
Term prioritization combines network support with scores obtained after randomly removing edges, rebuilding the ontology and aligning it to the original. The printed bootstrap expression is the reciprocal of n times the sum of alignment scores. It is not the ordinary arithmetic mean and becomes undefined if that sum is zero. The intended formula and author implementation were not verified, so this record does not silently replace it or adopt a numerical robustness guarantee.
Scope
#evaluation-and-input-dependence
The study uses yeast interaction networks and comparisons with prepared GO hierarchies, functional enrichment and selected genetic-interaction profiles. It acknowledges GO-dependent thresholds for expression and YeastNet inputs and reports controls removing those sources. Alignment, enrichment and selected Golgi experiments provide different evidence; none establishes correctness of every inferred term or relation. Supplementary controls were described in the main text but were not independently inspected here.
⚠️ Curated Limitations & Negative Results
- The record denotes the 2013 construction algorithm, not the released yeast ontology, NeXO Web database or a generic ontology reasoner. The associated alignment algorithm is a separate operation.
- HAC-ML supplies relevant upstream method context, but the displayed Beta objective and optimization terminology remain distinct from an inspected implementation. Tree pruning and added-parent ordering can affect the result.
- The second-parent criteria and density thresholds are source-specific heuristics. Gene-set membership and cross-interaction counting after earlier additions need explicit implementation conventions.
- The published and advance PDFs show the same reciprocal bootstrap expression on page 10. Extracted-text comparison found only publication-header, pagination and copyright differences; this is not a complete visual audit of both versions.
- The inspected published PDF has pre-existing highlights. All ten pages, references, five figures and method equations were read; supplementary datasets and figures remain uninspected.
- No author implementation, experimental dataset or supplementary archive was audited or executed. Finite curator calculations assess printed formulas and bounded examples, not benchmark replication or global optimality. The record remains DRAFT.
Artificial Intelligence Ontology (AIO) Grounding
Mapped AIO Classes
| AIO Target Class | Mapping Relation | Mapping Rationale |
|---|---|---|
AIO:HierarchicalClustering
Hierarchical Clustering
|
related_match | The algorithm constructs a hierarchy of object groups. The target is a related task, so this mapping does not assert algorithm/task equivalence or subclass membership. |
AIO:UnsupervisedClustering
Unsupervised Clustering
|
related_match | The grouping procedure uses network relationships rather than training a classifier on object labels. Domain-informed input selection or GO-dependent evaluation does not establish identity with a supervised classifier. |
Evidence Sources & Primary Literature
2 Sources| Source ID | Publication / Document | Type / Access | Reference & Checksum |
|---|---|---|---|
dutkowski2013
|
Janusz Dutkowski, Michael Kramer, Michal A Surma, Rama Balakrishnan, J Michael Cherry, Nevan J Krogan, Trey Ideker
|
primary paper
full_text
|
ID: doi:10.1038/nbt.2463
sha256:77571ca2fa83…
|
park2011
|
Yongjin Park, Joel S Bader
|
primary paper
full_text
|
ID: doi:10.1186/1471-2105-12-S1-S44
sha256:e930abd762f0…
|
Curation Provenance
Curator: Codex
Method: ai_assisted
Model: GPT-6
Created: 2026-10-14T01:04:45Z
Updated: 2026-10-14T01:04:45Z
Source File:
data/curated/NeXOOntologyConstruction.yaml