Within-species gene-content ecotypes are groups of genomes clustered by accessory gene-content differences. Comparing their COG (Clusters of Orthologous Groups) profiles shows that ecotypes differ in both adaptive and housekeeping functions, with larger proportional shifts in adaptive categories. 1
This result also illustrates why statistical significance and effect size answer different questions: a statistical test evaluates whether an observation is unlikely under a specified null hypothesis, whereas an effect size describes the magnitude or biological consequence of the difference. 1 The evidence therefore supports non-random functional differentiation, but does not imply that every significant category has a large functional consequence. 1
Key Evidence
The adaptive set comprised COG categories V (defense), P (inorganic ion transport and metabolism), G (carbohydrate transport and metabolism), E (amino acid transport and metabolism), Q (secondary metabolites biosynthesis, transport, and catabolism), M (cell wall, membrane, and envelope biogenesis), and K (transcription). The housekeeping set comprised J (translation), F (nucleotide transport and metabolism), H (coenzyme transport and metabolism), and C (energy production and conversion). 1 The report treats the prespecified hypothesis that these adaptive categories (V, P, G, E, Q, M, K) would differentiate more than these housekeeping categories (J, F, H, C) as only partially supported. 1
Across 257 chi-square or Fisher’s exact tests covering 12 species and 23 COG categories, 170 tests (66.1%) were significant after Benjamini–Hochberg false-discovery-rate (BH-FDR) correction at q < 0.05; all 12 species with valid clusters had at least one significantly differentiated category. 1 Adaptive categories were significant in 67 of 84 tests (79.8%), compared with 33 of 48 tests (68.8%) for housekeeping categories, a 1.16x significance-rate ratio. 1
The effect-size contrast was stronger: mean effect size was 0.0136 for adaptive categories and 0.0064 for housekeeping categories, a 2.13x ratio. A one-sided Mann–Whitney U test comparing the distributions gave p = 2.53 x 10^-6. 1 Thus, the results support stronger adaptive differentiation in magnitude while refining any claim of absolute adaptive-versus-housekeeping separation: housekeeping categories differentiated frequently and had a 68.8% significance rate. 1
The most frequently differentiated categories were E (amino acid metabolism), significant in 11/12 species (91.7%, rate 0.917); S (unknown function), significant in 11/12 species (91.7%, rate 0.917); V (defense), significant in 11/12 species (91.7%, rate 0.917); and G (carbohydrate metabolism), significant in 10/12 species (83.3%, rate 0.833). 1 Mean effect sizes for those four categories were 0.0120 for E, 0.0392 for S, 0.0078 for V, and 0.0176 for G. 1 The least differentiated categories were A (RNA processing), significant in 1/9 tested species; B (chromatin), significant in 1/8; and Z (cytoskeleton), significant in 2/6. 1 Categories M, L, C, P, and Q were each significant in 9/12 species (rate 0.750), with mean effect sizes of 0.0155 for M, 0.0337 for L, 0.0095 for C, 0.0170 for P, and 0.0112 for Q, while K, J, F, and H were each significant in 8/12 species (rate 0.667), with mean effect sizes of 0.0141 for K, 0.0051 for J, 0.0041 for F, and 0.0069 for H. 1 Adaptive-category mean effect sizes were 0.0078 for V, 0.0176 for G, 0.0155 for M, 0.0170 for P, 0.0112 for Q, 0.0141 for K, and 0.0120 for E. 1 Housekeeping categories J, F, and H were each significant in 8/12 species, while C was significant in 9/12 species; their mean effect sizes were 0.0051 for J, 0.0041 for F, 0.0069 for H, and 0.0095 for C. 1
The largest mean effect sizes overall were 0.0392 for S (unknown function), significant in 11/12 species, and 0.0337 for L (replication, recombination, and repair), significant in 9/12 species, showing that the strongest differences were not confined to the predefined adaptive set. 1 Reported at lower precision as S: 0.039 and L: 0.034, these largest mean effects correspond to “~3-4 percentage point differences in COG category proportions between ecotypes”, which the report calls statistically significant because of large sample sizes but modest in biological magnitude. 1 The large S effect is itself read as an annotation gap: substantial ecotype differentiation involves genes of currently unknown function. 1 The report characterizes these effect sizes as small and notes that significance may partly reflect the large sample size: 1,820 genomes were assigned across 12 species. 1
Interpretation and Boundaries
The findings are consistent with a model in which ecotype-defining gene-content differences preferentially alter defense, transport, secondary metabolism, carbohydrate use, amino-acid metabolism, and envelope functions while also affecting core cellular processes. 1 Widespread significance is therefore not evidence that all differences are large, and the significance-versus-effect-size comparison supports reporting both quantities together. 1 Published work on Sulfurovum ecotypes at hydrothermal vents reported COG categories P and S as the most variable between ecotypes; the present result, with P significant in 9/12 species and S in 11/12 species and carrying the largest effect, is concordant and extends the comparison from a single deep-sea species to 12 diverse species, with the earlier study entering here only as secondary literature evidence. 1
Evidence from laboratory fitness and gene conservation refines this contrast: core genes were more burdensome than non-core genes in Protein Metabolism, Motility, and RNA Metabolism, whereas non-core genes were more burdensome in the Cell Wall category. 2 The same analysis identified 25,271 condition-dependent trade-off genes and 28,017 genes that were both costly and conserved, suggesting that conservation may preserve functions whose laboratory costs are offset by benefits in other environments. 2 Across approximately 194,000 genes from 43 bacteria, essential genes were 82% core whereas always-neutral genes were 66% core, although the association was weak overall. 3 Strong condition-specific effects were enriched among core genes (77.3% core versus 70.3% without specific phenotypes), and core genes had heavier fitness-effect tails in both negative and positive directions. 3 These findings support condition-dependent interpretation rather than an assumption that housekeeping-associated or conserved functions are uniformly low-cost or constitutive. 2,3
The broader 32-species COG comparison supports the functional partitioning but refines its scope: novel or singleton genes were enriched in defense category V by +2.83%, while the categories that report lists as consistently core-enriched are translation J at -4.65% enrichment with 97% consistency (a value the same report also labels the strongest depletion), nucleotide metabolism F at -2.09%, coenzyme metabolism H at -2.06%, amino acid metabolism E at -1.81%, and energy production C at -1.75%; the signed values are carried here exactly as given, because the source’s own enrichment-versus-depletion wording does not fix a direction for them. Novel or singleton genes were also enriched in mobile-element category L by +10.88%, with 100% consistency across species, and in V by +2.83%, with 100% consistency. These patterns spanned 9 phyla and 357,623 genes, but describe core-versus-novel classes rather than within-species ecotype contrasts. 4
The comparison assigned 1,820 genomes across 12 species spanning 6 phyla, with valid gene-content ecotypes in 12 of 15 sampled species (80%); the sample came from 456 eligible species and may not represent broader phylogenetic or ecological diversity. 1 Approximately 38% of gene clusters had COG annotations, leaving 62% unannotated, so the measured effects describe annotated COG space; the report warns that the 62% without annotations may include many ecotype-specific adaptive genes in poorly characterized functional categories, potentially biasing results toward well-annotated housekeeping functions. 1 The broader comparison reported approximately 70% gene annotation coverage; this refines rather than replaces the caveat, since the differing figures may reflect different datasets or annotation denominators. 4 Singleton neutrality may additionally reflect poor transposon coverage rather than true absence of function. 3
The observed association is between gene-content clusters and functional-category proportions, not direct evidence that adaptive categories cause ecological differentiation. 1 Across 172 species, the median partial correlation between environmental similarity and gene-content similarity was 0.0025, compared with 0.0143 for phylogeny; phylogeny dominated in 60.5% of species, and environmental effects were non-significant in 156 species (90.7%). 5 This does not negate the larger adaptive-category effect sizes but suggests the hypothesis that ecological differentiation is concentrated in specific gene subsets rather than governing whole-genome gene-content similarity. 5 Environmental and host-associated bacteria did not differ significantly in environmental effects (p=0.66), and AlphaEarth embeddings covered only 28.4% of genomes. 5
Without within-species phylogenetic controls such as core-genome trees, the analysis cannot distinguish ecological adaptation from phylogenetic or demographic substructure. 1 A small but significant category shift should therefore not automatically be treated as an independently evolved ecological adaptation; this supports linking the interpretation to ecotype-environment-gene-content, phylogenetic-confounding-of-pangenome-associations, ontology-and-category-schema-sensitivity, genomic-under-representation, and phenotype-database-coverage-bias. 1
Cross-System Replication in Other Ontologies
An independent test in human gut commensals supports the direction of this contrast outside the COG scheme and outside within-species ecotype clustering. Across 23,579 strain-adaptation genes at FDR (false-discovery rate) < 0.10 in 59 species — 12,159 biased toward inflammatory bowel disease (IBD) and 11,420 health-biased — the IBD-biased gene set was 1.38x enriched for adaptation functions (Fisher odds ratio, OR = 1.38, p = 2.4e-6) and 0.62x depleted for housekeeping functions (OR = 0.62, p = 6.4e-20), with the listed adaptation themes being mucin/glycan utilization, two-component signaling, antibiotic resistance, and virulence; the strict housekeeping-domination falsifiability bound was rejected. 6
The underlying proportions were 2.13% adaptation among IBD-biased genes versus 1.59% background (7 subcategories) and 2.81% housekeeping versus 4.36% background (6 subcategories), with 95.06% of IBD-biased genes and 94.05% of background genes unclassified. 6 The report judges the corresponding hypothesis (H3b) supported, because the IBD-biased set was enriched for classified adaptation genes and depleted for housekeeping genes, concluding that the gene-content signal is statistically distinct from housekeeping artifact and consistent with a biologically real IBD-niche gene signature operating across multiple commensal species. 6 Because 95.06% of IBD-biased genes and 94.05% of background genes are unclassified, this enrichment describes classified gene space only — the same annotation-coverage boundary that limits the COG-based comparison above. 6
Within the classified IBD-biased adaptation genes, mucin/glycan utilization was the largest subcategory with 94 genes, followed by two-component signaling (52), antibiotic resistance (35), membrane transport specialty (33), virulence/secretion (15), bile acid (15), and oxidative stress response (15); mucin/glycan utilization is reported as the dominant IBD-adaptation theme. 6 This refines the COG-based picture by naming concrete adaptive functions — host-glycan foraging, signaling, efflux — in place of broad transport and metabolism categories. 6
The cross-species shared signal is described as dominated by transport, regulation, stress response, and DNA repair: KEGG orthologs K03088 (sigma factor RpoE/RpoS, environmental stress response) IBD-biased in 21 species, K01990 (ABC transporter ATP-binding protein) in 20, K06147 (ABC-2 type transport system) in 19, K02529 (LacI/GalR family transcriptional regulator) in 18, K07240 (TIM44-like protein) in 17, K03424 (TatD-related deoxyribonuclease) in 17, K08303 (SmpB, tmRNA-binding, translation rescue) in 16, and K03466 (DNA topoisomerase) in 15, together with gene symbols corC (Mg2+/Co2+ efflux), luxQ (autoinducer-2 quorum sensing), pglA and rfbC (capsule polysaccharide / LPS), bepC (efflux pump), and abgB (aminobenzoyl-glutamate utilization), each consistent across 9-10 species. 6 The report characterizes this as a classic niche-adaptation signature; since translation rescue and topoisomerase activity would conventionally be grouped with core cellular machinery, this suggests the hypothesis that the adaptation/housekeeping boundary drawn by that project’s gene-symbol classifier differs from the COG split used here, but the source does not state how those particular orthologs were classified, so the difference is untested — an ontology dependence shared with ontology-and-category-schema-sensitivity. 6
The same project also shows adaptive capacity distributed unevenly across lineages rather than evenly over species: Fisher enrichment of biosynthetic gene cluster (BGC, a genomic cluster encoding one secondary-metabolite pathway) themes gave iron_siderophore OR = 44.4 (FDR 6e-56), genotoxin_microcin OR = 234 (FDR 3e-35), and NRPS-PKS-hybrid (nonribosomal peptide synthetase / polyketide synthase hybrid) OR = 1.5 (FDR 0.04), and escherichia-coli alone among the report’s Tier-A core species carries the iron-plus-genotoxin BGC signature, with 54 iron MIBiG (Minimum Information about a Biosynthetic Gene cluster database) entries — 19 Yersiniabactin + 16 Enterobactin + 19 siderophore-class — alongside 25 genotoxin entries. 6 The project’s summary gives the genotoxin breakdown as 8 Colibactin + 15 Microcin B17, which is incomplete: the detailed results state that the 25 genotoxin BGCs comprise 8 Colibactin, 15 Microcin B17, and 2 Microcin J25. 6 Expression of the ebf and ecf gene sets, measured as RPKM (reads per kilobase of transcript per million mapped reads), was higher in Crohn’s disease (CD) across 4 cohorts; the summary states this jointly as p<1e-31, whereas the detailed table reports meta p = 1.1e-31 for ebf and meta p = 5.1e-33 for ecf, so the single summary threshold should not be read as either per-gene-set value. 6 Concentration of the strongest adaptive-function enrichments in a single species refines the ecotype-level result: large adaptive-category shifts need not be a general property of all lineages — see pathobiont-biosynthetic-gene-cluster-repertoires and iron acquisition and pathobiont expansion. 6
These gut results carry boundaries that limit how far the replication framing can be pushed: the H3b test used one cohort only (Kumbhari/LSS-PRISM), and the planned cross-cohort sensitivity test (NB10b) was specified as sensitivity-only and is not executed in the project, so cohort portability is untested. 6 In the same analysis F. plautii yielded zero FDR<0.10 strain-adaptation genes out of 3,245 genes tested, an informative null that the report reads as CD-association operating through species-level abundance rather than strain-level genomic adaptation. 6 A single-cohort enrichment measured over a classified minority of genes therefore supports the adaptive-over-housekeeping direction without establishing it as a general property of gut commensal pangenomes. 6
In plant-associated bacteria, by contrast, a putatively adaptive function class sits predominantly in the core genome: beneficial plant-growth-promoting (PGP) gene clusters had a 64.6% core fraction, significantly exceeding both the genome-wide baseline of 46.8% and pathogenic gene clusters at 45.2% core fraction, with a Mann-Whitney U test on per-species core fractions giving U = 83,567,419 (p = 3.38e-125) and a bootstrap 95% confidence interval for the beneficial-pathogenic difference of [0.089, 0.106]. 7 This directly measured and strongly supported result refines the adaptive-versus-housekeeping framing by separating two axes that are easily conflated: which functional categories differentiate, and whether the differentiating genes are core or accessory. 7
The dominant functional themes among plant-enriched orthologous groups (OGs, clusters of homologous genes across genomes) were electron transport, iron-sulfur biosynthesis, and ion transport, including the high-affinity cytochrome oxidases COG1845 and COG0843; the source reads these as supporting enhanced metabolic competitiveness in the rhizosphere rather than plant-specific signaling, which is an interpretation rather than a measured mechanism. 7 Two of those themes fall on opposite sides of the split used here — inorganic ion transport (P, adaptive) and energy production (C, housekeeping) — so habitat-differentiating signal is not confined to the adaptive set. 7,1
Tensions
Is adaptive-function differentiation an accessory-genome phenomenon? One side: the broader cross-species comparison places adaptive-leaning categories in the novel or singleton gene class — mobile elements L at +10.88% enrichment with 100% consistency and defense V at +2.83% with 100% consistency — while listing housekeeping-leaning categories J, F, H, E and C with the core-gene class, at the signed values recorded above. 4 The other side: beneficial plant-growth-promoting gene clusters had a 64.6% core fraction against 45.2% for pathogenic clusters and a 46.8% genome-wide baseline, placing an adaptive marker class inside the core. 7 The two sides use different function definitions (COG category membership versus curated beneficial/pathogenic markers) and different units (gene classes versus marker clusters), so the disagreement is partly about what counts as an adaptive function; it must not be settled by averaging the core fractions. Resolving work would score the same genome set under both schemes and compare core fractions category by category, and test whether the categories with larger ecotype effect sizes are the accessory-leaning ones — connecting this page to two-speed-bacterial-genome and dual-nature-plant-associated-bacteria. 4,7
Practical Reporting Principle
Comparative analyses should report the number of tests, correction procedure, exact significance threshold, proportion of significant results, effect-size distribution, and sample size together. 1 Here, the combination of 170 significant results out of 257 tests, mean effect sizes of 0.0136 and 0.0064, the 1,820-genome sample, and the stated annotation and phylogenetic limitations is more informative than the significance count alone. 1
The underlying project summaries are Ecotype Functional Differentiation, COG Functional Category Analysis, Ecotype Correlation Analysis, Core Gene Paradox — Why Are Core Genes More Burdensome?, Fitness Effects vs Conservation — Quantitative Analysis, Metagenome-Prioritized Phage Cocktails for Crohn’s Disease and IBD, and Plant Microbiome Ecotypes — Compartment-Specific Functional Guilds and Their Genetic Architecture.
Open Directions
- Use the 1,820 genome-to-ecotype assignments and core-genome phylogenies with phylogenetically controlled models, confidence intervals, or bootstrap resampling to test whether the 2.13x adaptive-versus-housekeeping mean-effect-size ratio and the small effects remain stable after accounting for shared ancestry. 1
- Extend the analysis from the 15-species sample to all 456 eligible species, modeling effect size against genome count and testing whether the 79.8% versus 68.8% significance-rate contrast persists across genome-count bins and phylogenetic groups. 1
- Combine ecotype assignments with habitat metadata, alternative environmental distances, and multivariate association methods to test whether specific COG shifts predict environmental differences rather than lineage structure. 1,5
- Analyze the 62% of unannotated gene clusters with AlphaFold or domain analysis and test whether their inclusion changes the adaptive-versus-housekeeping effect-size contrast. 1
- Recluster the same gene-content data with HDBSCAN and compare category-level significance and effect sizes with KMeans to test dependence on spherical-cluster assumptions and the selected value of k. 1
- Separate mobile-element and defense categories in the broader COG comparison and test whether the +10.88% COG L and +2.83% COG V novelty enrichments predict within-species ecotype differentiation. 4
- Overlay ecotype-specific COG shifts with condition-resolved fitness measurements, controlling for transposon coverage, to test whether larger adaptive-category effects contain more costly conserved or condition-dependent trade-off genes and whether adaptive and housekeeping categories differ in fitness-effect breadth. 1,2,3
- Re-score the 23,579 FDR<0.10 gut strain-adaptation genes under the COG scheme used here, and the 1,820 ecotype-assigned genomes under the adaptation/housekeeping subcategories used there, to test whether the OR = 1.38 / OR = 0.62 pattern and the 2.13x mean-effect-size ratio survive a change of ontology rather than reflecting scheme-specific category boundaries. 1,6
- Repeat the single-cohort gut enrichment test on an independent cohort, since the planned cross-cohort sensitivity notebook was not executed, to close the gap over whether the adaptation-enrichment / housekeeping-depletion pattern is cohort-portable. 6
- Compute per-COG-category core fractions for the 1,820 ecotype-assigned genomes and compare them with the 64.6% beneficial and 45.2% pathogenic core fractions, closing the gap over whether the categories with the largest ecotype effect sizes are the accessory-leaning ones. 1,7