Motif Enrichment Table¶
Reporting artifact — instantiates Network Motif and Pattern Discovery
Lays observed against expected motif counts with effect size, uncertainty, and multiple-comparison control, turning a pile of counts into a defensible enrichment verdict and a cross-network profile.
Raw counts do not settle anything: a motif that appears four hundred times may be exactly what chance predicts, and one appearing nine times may be a genuine surprise. Motif Enrichment Table is the artifact that adjudicates. For each candidate motif it sets the observed count beside the count expected under a null, then computes the things that actually decide whether "enriched" is warranted — an effect size, an uncertainty interval, and a correction for the fact that many motifs were tested at once. Its distinctive contribution is the significance verdict, laid out auditable row by row. It does not produce the observed counts and it does not generate the null; it consumes both. What it adds is the judgment layer — and, when several networks are tabulated side by side, a profile that lets whole networks be compared and classified by their motif signatures rather than by any single pattern.
Example¶
An ecologist has counted three-species subgraphs in a set of food webs — species as nodes, "eats" as directed edges — and wants to know which trophic motifs are genuinely characteristic of one lake's web. She assembles the enrichment table. Each row is a three-species motif: the observed count in the real web, the mean and spread of counts across a null ensemble of degree-matched webs, a z-score-style effect size, and a p-value. The tri-trophic food chain (A eats B eats C) shows an observed count well above the null mean; apparent enrichment. But the table also flags that thirteen motifs were tested, so it applies a false-discovery-rate correction; after it, two of the "significant" motifs fall away as expected under multiple testing, and only the food chain and one omnivory triangle survive with both a large effect and a corrected p below threshold.
She then widens the table into a panel: the same rows, one column per lake. The pattern of which motifs are over- and under-represented — the web's significance profile — turns out to cluster the lakes into two groups, separating webs with strong omnivory from strictly hierarchical ones. The individual counts never told that story; the enrichment table, corrected and profiled, did.
How it works¶
The table's distinguishing work is turning two count vectors — observed and expected — into a defensible verdict:
- Compute an effect size, not just a test. For each motif, express how far the observed count sits from the null's center in units of the null's spread, so magnitude is visible independent of network size.
- Attach uncertainty. Carry the null distribution's variance and a confidence or credible interval, so a large point estimate on a noisy null is not mistaken for a solid one.
- Correct for multiplicity. Because many motifs are tested together, adjust p-values or control the false-discovery rate; an uncorrected table manufactures false positives by sheer count of tests.[n1]
- Filter on practical support. Drop motifs below a minimum instance count, where enrichment ratios are unstable, before ranking.
- Widen to a panel. Add one column per network to read motif profiles across systems.
Tuning parameters¶
- Effect-size metric — z-score, log-ratio, or standardized subgraph ratio. Each weights large-count and rare motifs differently.
- Multiple-comparison method — Bonferroni (strict, few false positives, many missed), false-discovery-rate control (permissive, tolerates a known false fraction), or none (indefensible at scale).
- Significance threshold — the effect-size and corrected-p cutoffs a motif must clear to be called enriched.
- Minimum support — the instance floor below which a motif is excluded as too rare to score stably.
- Normalization for the panel — whether cross-network profiles are compared on raw effect sizes or on length-normalized profiles, which controls for network size when classifying.
When it helps, and when it misleads¶
Its strength is that it separates frequent from enriched and makes the separation auditable: every claim of over-representation carries its effect size, its uncertainty, and its multiplicity correction in the same row, and the panel form turns a set of networks into a comparable profile that can reveal superfamilies sharing a motif signature. It is the discipline that keeps "this motif is common" from being mistaken for "this motif matters."
Its failure mode is the seductiveness of a significant number in a large graph. Statistical significance is nearly free when counts are huge — a trivially small enrichment can clear any p-threshold — so a table that ranks on p-value alone will elevate motifs of no practical import. The classic misuse is running the table against the wrong null (a naive random graph rather than a degree-matched one) and then trusting its confident p-values, laundering a null-model mismatch into decimal-place authority. The guarding discipline is to rank on effect size and support first and significance second, to state which null the "expected" column came from, and to treat the table as an argument whose inputs must hold — not a verdict.
How it implements the components¶
Motif Enrichment Table fills the significance-and-comparison layer:
significance_and_effect_filter— it computes effect sizes, uncertainty, multiplicity correction, and minimum-support filtering, combining them into the enriched/not verdict the archetype demands.cross_network_comparison_panel— in its multi-column form it aligns motif profiles across networks, enabling comparison and classification by motif signature.
It does not enumerate subgraphs or produce the observed counts (subgraph_enumeration_process, recurrence_measurement) — those come from Subgraph Census; and it does not generate the expected counts it compares against (baseline_or_null_model) — that is Random Graph Null Ensemble. The table consumes both and judges them.
Related¶
- Instantiates: Network Motif and Pattern Discovery — the reporting step that converts counts into a defensible enrichment claim.
- Consumes: Subgraph Census for observed counts and Random Graph Null Ensemble for expected counts and their spread.
- Sibling mechanisms: Subgraph Census · Random Graph Null Ensemble · Degree-Preserving Edge Swap · Graph Motif Mining Algorithm · Motif Role Hypothesis Card · Domain Expert Motif Review
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Motif Enrichment Table operates as a computation, comparison, model, or analytic representation used to infer, estimate, or choose because it lays observed against expected motif counts with effect size, uncertainty, and multiple-comparison control, turning a pile of counts into a defensible enrichment verdict and a cross-network profile.
Independent corroboration: The frozen evidence defines Motif Enrichment Table as 'Lays observed against expected motif counts with effect size, uncertainty, and multiple-comparison control, turning a pile of counts into a defensible enrichment verdict and a cross-network profile', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Systems Thinking & Cybernetics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Network-motif research directly combines recurring substructures with enrichment tests in complex systems; statistics, ecology, and algorithms are co-formative. This establishes systems_cybernetics as the primary origin lineage rather than merely a domain where the mechanism is now applied.
Related originating lineages:
- Biology & Ecology — Food-web and systems-biology applications materially shaped motif interpretation across living networks.
- Computer Science & Software Engineering — Graph algorithms supply subgraph enumeration and network comparison.
- Statistics & Experimental Design — Observed-versus-expected enrichment with uncertainty, effect size, and multiplicity control is fundamentally a statistical inference artifact.
Review resolution: Authoritative/primary-source research resolves the conflicting primary-origin claims in favor of systems_cybernetics: Network-motif research directly combines recurring substructures with enrichment tests in complex systems; statistics, ecology, and algorithms are co-formative. Retained alternate origins (statistics_experimental_design, biology_ecology, computer_science) are limited to independently formative or materially shaping lineages supported by the reviewer evidence; downstream adoption alone was not promoted to origin. The breadth of present-day use is recorded separately as domain_reach=multi_domain. origin_mode=cross_disciplinary_synthesis, confidence=medium, and encyclopedia_synthesis=false reflect the surviving provenance evidence and the encyclopedia's generalization.
Review outcome: Researched adjudication after independent review; medium confidence.
Sources consulted:
- Network motifs: simple building blocks of complex networks — Establishes motif detection, enrichment, and functional-role hypotheses in systems/network science.
Notes¶
[n1] When many hypotheses are tested at once, controlling the false-discovery rate — the expected fraction of declared discoveries that are false, as in the Benjamini–Hochberg procedure — is the standard correction; without it, testing dozens of motifs guarantees some will cross a naive significance line by chance alone. ↩