Costly, non-conserved genes are a distinct pangenomic class whose laboratory burden and limited distribution suggest ongoing evolutionary instability rather than persistent core metabolism. 1 The analysis examined 142,190 genes from 43 bacteria and identified 5,526 genes in this class. 1 The underlying evidence and methods are summarized in The 5,526 Costly + Dispensable Genes. 1 The report defines the class as genes that are burdensome and not conserved in the pangenome at the same time. Burdensome means that fitness improves when the gene is deleted. The report presents this as the first systematic characterization of such genes. 1
Central claim
The report supports the interpretation that many costly+dispensable genes are horizontally acquired genomic material—including insertion sequences, prophage remnants, transposases, integrases, phage sequences, recombinases, defense systems, and plasmid-associated DNA—that imposes a fitness burden and is subsequently vulnerable to gene loss. 1 The report describes the 5,526 costly+dispensable genes as overwhelmingly associated with mobile genetic elements and calls them the genomic debris of horizontal gene transfer. That origin is the authors’ reading of annotation and distribution patterns, not a direct measurement of where the genes came from. 1 This interpretation refines gene-essentiality by separating laboratory-measured burden from long-term pangenome conservation: a gene can reduce laboratory fitness without being evolutionarily retained across strains. 1
The evidence is strongest for mobile and recently acquired DNA, but the report does not establish that every costly+dispensable gene is nonfunctional or destined for deletion. 1 Instead, the findings suggest the hypothesis that selection removes costly accessory genes unless environmental or community conditions periodically reward their functions. 1 New evidence that trade-off genes are 1.29 times more likely to be core than non-core (odds ratio 1.29, p=1.2e-44) refines this interpretation: costly, condition-dependent genes are not uniformly unstable, and some are retained broadly because their costs and benefits vary across conditions. 2
The cross-project synthesis supports the costly-versus-dispensable selection signature: it identifies the 5,526 costly-and-dispensable genes as candidates for ongoing gene loss, while treating 28,017 genes that are both costly in the laboratory and conserved as stronger evidence for retention under purifying selection in environments not represented by the assay. 3 This distinction is evidence for, rather than a direct measurement of, natural selection and therefore supports the cautious interpretation above. 3 The synthesis itself leaves the identity of the 5,526 costly+dispensable genes unresolved. It only poses mobile elements and recently acquired genes as possibilities. 3 The central discoveries digest repeats the same matrix. It identifies 28,017 genes as costly in the laboratory yet conserved in the pangenome, 5,526 as costly and dispensable (candidates for ongoing gene loss), and 21,886 as neutral and dispensable (niche-specific). 4 The digest says every strain keeps the costly conserved genes because nature requires them in conditions the laboratory does not capture. That statement extrapolates from laboratory fitness and is not a direct measurement of selection in natural environments. 4
The broader fitness–conservation analysis refines this class-based interpretation: across approximately 194,000 genes from 43 bacteria, essential genes were 82% core whereas always-neutral genes were 66% core, but breadth of fitness effects was only weakly associated with conservation (Spearman rho=0.086, p=8.1e-230). 5 In a separate integration of Fitness Browser genes with pangenome clusters, essential genes were 86.1% core versus 81.2% for non-essential genes, with a median odds ratio of 1.56 across 33 organisms; 18 of 33 organisms showed significant enrichment after Benjamini-Hochberg false discovery rate correction. 6 This supports the direction of the fitness–conservation association while refining its interpretation: essentiality is associated with only modestly greater core conservation, so costly+dispensable genes remain a specific candidate class for loss rather than evidence that laboratory fitness alone predicts evolutionary fate. 5,6
Mobile-element-associated burden
Costly+dispensable genes are 7.45 times more likely than costly+conserved genes to contain mobile-element keywords such as transposase, integrase, phage, IS element, recombinase, or prophage (OR=7.45, p=4.6e-71). 1 The SEED category “Phages, Prophages, Transposable elements, Plasmids” is enriched 11.7-fold in the costly+dispensable group (FDR=1.3e-17), where FDR means false discovery rate. 1 “Virulence” is enriched 26.7-fold (FDR=5.6e-14), although that result is based on small counts of 21 versus 4 genes. 1
These enrichments support the interpretation that a measurable fraction of laboratory fitness burden is associated with mobile or recently mobile DNA rather than conserved metabolic machinery, and refine horizontal-gene-transfer-driven-innovation by indicating that horizontal acquisition can produce genes that are potentially useful in particular settings yet burdensome in laboratory conditions. 1 The inference is strongest for the enrichment statistics: the analysis used annotation keywords and functional categories, not direct measurements of element mobility or transfer, and therefore does not distinguish active transposable elements from prophage remnants, inactive fragments, defense systems, or other recently acquired genes. 1
Separate evidence from antimicrobial-resistance (AMR) genes supports the idea that specific adaptive functions can carry a relative laboratory burden. Under non-antibiotic conditions, AMR gene knockouts show systematically higher fitness than non-AMR gene knockouts. A DerSimonian-Laird random-effects meta-analysis across 25 organisms yields a pooled effect of +0.086 [95% CI: +0.074, +0.098], z = 14.3, p ~ 0. 7 The report says this effect means AMR gene knockouts are less detrimental than the average gene knockout. AMR knockouts still grow slower than the pool average, so the cost appears only as a relative difference from the non-AMR background. The result therefore does not directly show a fitness gain from deleting resistance genes. It also measures relative burden rather than pangenome conservation class, so it does not show on its own that resistance genes are lost. The topic is developed further in antimicrobial-resistance-fitness-cost. 7
A field-ecology report cites Peng et al. (2022) for horizontal acquisition of heavy-metal resistance by Rhodanobacter strains at Oak Ridge. 8 It links that cited finding to a costly_dispensable_genes result that metal-resistance genes are enriched in the accessory genome. The link is asserted by that report rather than independently verified. It is consistent with reading accessory burden as horizontally acquired, but it does not test that reading. 8
Evidence for recent acquisition and weak retention
Only 50.8% of costly+dispensable genes have SEED annotations, compared with 74.9% of costly+conserved genes. 1 KEGG annotation rates are 20.0% and 42.7%, respectively. 1 The costly+dispensable group contains 44.5% orphan genes with no ortholog group, compared with 13.1% among costly+conserved genes. 1
The median ortholog breadth is 15 organisms for costly+dispensable genes and 31 organisms for costly+conserved genes (Mann–Whitney p=4.0e-99, rank-biserial r=0.233). 1 Costly+dispensable genes also have a 24.2% singleton fraction, whereas no costly+conserved genes are singletons; this comparison is partly structural because core genes cannot be singletons by definition. 1 Within the dispensable category, costly genes are only slightly more likely than neutral genes to be singletons (OR=1.09, p=0.02). 1
The costly+dispensable genes have a median length of 615 bp, compared with 765 bp for costly+conserved genes (p=4.2e-75, rank-biserial r=0.170). 1 Their shorter length is consistent with the presence of IS elements and gene fragments, but does not by itself prove that individual genes are degraded remnants. 1 Together, poor annotation, orphan status, narrow ortholog breadth, singleton frequency, and short length support gene-function-acquisition-depth and genomic-under-representation as relevant frameworks for distinguishing recent acquisition from deeply conserved cellular function. 1
The fitness–conservation analysis refines this interpretation by showing that novel singleton genes had near-zero mean fitness in the tested laboratory assays, while noting that poor transposon coverage may also produce this apparent neutrality. 5 This supports treating weak retention and measured neutrality as distinct observations rather than assuming that singleton genes are intrinsically unimportant. 5 The new pangenome linkage analysis further supports the functional-dark-matter interpretation for a different conservation class: 44.7% of essential-unmapped genes and 38.2% of essential-auxiliary genes were hypothetical, compared with 13.0% of essential-core genes. 6 Thus, poor annotation is not restricted to genes that are costly and dispensable; it can also mark essential genes whose sequence relationships or pangenome placement are unresolved. 6
Depletion from core metabolism
Fourteen SEED top-level categories are significantly depleted in costly+dispensable genes at FDR < 0.05, including Protein Metabolism, Respiration, Carbohydrates, Amino Acids, Cofactors/Vitamins, Motility, Stress Response, and RNA Metabolism. 1 By contrast, the 28,017 costly+conserved genes are maintained in categories that include core cellular functions such as Protein Metabolism, Respiration, and Motility. 1
This contrast supports core-genome-burden-paradox: laboratory cost does not imply evolutionary loss when natural selection can offset that cost in environments not represented by the laboratory assay. 1 The synthesis provides an aligned quantitative comparison: 24.4% of core genes show positive fitness when deleted, compared with 19.9% of accessory genes, and core genes are 1.78x more likely to have strong condition-specific phenotypes. 3 Thus, the lower retention of costly+dispensable genes does not mean that accessory genes are generally more burdensome; it supports the existing interpretation that burden and conservation interact differently for core and accessory material. 3 The synthesis records this as a null result for the streamlining hypothesis: accessory genes are not systematically burdensome and are less costly than core genes, not more. 3
The fitness–conservation analysis supports this paradox and extends it beyond the costly+conserved comparison: core genes had heavier fitness-effect tails in both negative and positive directions, with 24.4% ever beneficial when deleted versus 19.9% of auxiliary genes (OR=0.77 for auxiliary vs core). This runs counter to the expectation that accessory genes are more costly to carry. 5 Core genes can therefore combine broad retention with condition-dependent costs and benefits, rather than representing uniformly low-burden functions. 5
The core-gene trade-off analysis further supports this paradox, finding 25,271 trade-off genes overall (17.8% of genes examined) and showing that core genes are more likely than non-core genes to fall in this condition-dependent class. 2 However, the effect is function-specific: core genes are more burdensome in Protein Metabolism, Motility, and RNA Metabolism, whereas non-core genes are more burdensome in Cell Wall functions, with differences of +6.2, +7.8, +12.9, and -14.1 percentage points, respectively. 2 This refines the interpretation by showing that conservation does not impose one universal burden pattern across functional categories. 2
The comparison also supports pangenome-integration by showing that laboratory fitness measurements and pangenome conservation provide complementary evidence about gene persistence. 1 The 28,017 Costly + Conserved genes are interpreted as genes retained despite laboratory cost, whereas the report’s selection-signature matrix labels the 5,526 Costly + Dispensable genes as candidates for ongoing gene loss, not demonstrated losses. They remain the more specific class associated with weak retention and possible loss. 2 The new integration supports this complementarity by linking 177,863 Fitness Browser gene-to-cluster pairs, while also showing that pangenome coverage and gene-identifier compatibility can limit inference: 44 of 48 organisms were mapped, 33 entered downstream analysis, and ten were excluded for <90% DIAMOND coverage. 6
Conditional persistence
The costly+dispensable class is not phenotypically inert: 14.1% of its genes have condition-specific phenotypes, compared with 16.7% of costly+conserved genes and 2.7% of neutral+dispensable genes. 1 This result supports condition-specific-fitness and suggests the hypothesis that episodic benefits can delay the loss of genes that are costly under tested laboratory conditions. 1
The condition-specific result does not demonstrate that these genes are maintained by selection in nature because the phenotype data are biased toward conditions that can be tested in the laboratory. 1 A gene may therefore appear costly and dispensable in the available data while retaining value in unmeasured environments, hosts, or microbial communities. 1 The stronger core enrichment among trade-off genes supports this caution and indicates that condition dependence can characterize conserved genes as well as accessory genes. 2 The synthesis likewise states that laboratory measurements do not directly establish natural-environment fitness, reinforcing the need to treat evolutionary loss as a hypothesis rather than an observed fate. 3
The new analysis supports this caution: genes with strong condition-specific phenotypes were 77.3% core, versus 70.3% core among genes without specific phenotypes (OR=1.78, p=1.8e-97). These figures are core fractions within each phenotype class, not the share of core genes that show specific phenotypes. In addition, and 4,450 genes (2.7%) were neutral overall but critical in one condition; these ephemeral-niche genes were more common among core genes (3.0%) than auxiliary genes (1.7%) or singleton genes (1.6%). 5 This refines the proposed persistence mechanism by showing that conditional effects are not concentrated only in accessory genes. 5 Essentiality evidence also refines the condition argument: the new essentiality calls were inferred from RB-TnSeq, a random-barcode transposon sequencing approach, under library-construction and growth conditions represented in the Fitness Browser, so they do not capture genes essential only under untested stresses. 6
Adaptive loss and evolutionary lag
The costly_dispensable_genes report places its result alongside the Black Queen Hypothesis of Morris et al. (2012). Under that hypothesis, costly genes can be lost when community members provide their functions as public goods. 1 This connects gene loss to community-metabolic-interdependence. However, the report names the framework without testing whether community members provide these functions as public goods. 1 A separate analysis of latent metabolic pathways found a null result at the pathway level: conservation rates did not distinguish latent from active pathways. To explain that null, it proposes evolutionary lag. The Black Queen Hypothesis predicts ongoing gene loss, but the time frame over which loss is detectable may exceed what current genome sampling captures. Latent capabilities may therefore be mid-loss in some lineages yet still appear conserved at species level. 9 That proposal concerns latent pathways, not genes classified as costly by laboratory fitness. The report did not examine whether lag also applies to costly+conserved or costly+dispensable genes. Applying it to those classes would be an untested hypothesis, not a finding of the source. 9
The same report describes carbon source utilization pathways as 3.7× more likely to be latent than amino acid biosynthesis pathways (24.3% vs 6.5%). 9 From this it proposes a prediction it did not test: organisms in nutrient-rich environments should disproportionately lose carbon-pathway genes over evolutionary time, while amino acid biosynthesis genes are retained under stronger purifying selection. This is a testable loss prediction for latent pathways, not evidence that loss has occurred. 9
That report also calls for phylogenetic checks that apparent recent gene losses are real deletion events. By apparent losses it means singletons or auxiliary genes in the pangenome that fall within pathways classified as latent. 9 A similar longitudinal check is needed before costly+dispensable genes can be treated as being lost rather than as candidates for loss. 9
Organism-level heterogeneity
Pseudomonas stutzeri RCH2 contributes 21.5% of its genes as costly+dispensable, compared with 14.0% for the next organism, Bacteroides thetaiotaomicron. 1 The report identifies a recent mobile-element invasion or strain-specific genomic expansion as possible explanations, but leaves the cause unresolved. 1 This outlier qualifies the cross-organism interpretation and links the concept to subsurface-bacillota-specialization and horizontal-gene-transfer-driven-innovation, without establishing that either process caused the RCH2 pattern. 1
Interpretation and limits
The observed combination of MGE enrichment, narrow taxonomic distribution, short length, poor annotation, and depletion from core metabolism supports the hypothesis that many costly+dispensable genes are genomic debris or unstable accessory DNA associated with horizontal gene transfer. 1 The analysis does not establish that every such gene is degraded or inactive, nor does it by itself distinguish active mobility from remnants, defense systems, or other recently acquired genes. 1 The authors further describe costly+dispensable genes as almost exclusively mobile element debris rather than degrading metabolic pathways. They propose that genome expansion through horizontal gene transfer is the primary source of costly non-conserved genes in bacteria. Both statements are interpretation, not a direct test of gene origin. 1
“Costly” is defined as max_fit > 1 in any experiment, so classification can be driven by a single experiment and is sensitive to fitness-data noise. 1 The pangenome classification is binary, so it does not capture the fraction of genomes carrying each gene. 1 SEED and KEGG annotations cover only 56-79% of genes, and the unannotated fraction may have different functional profiles. 1 The 90% identity DIAMOND threshold used to link Fitness Browser measurements with pangenome genes may miss recently acquired genes with low sequence similarity. 1 Ortholog groups were assigned using bidirectional best hits across 48 organisms, so genes with orthologs outside that set may be misclassified as orphans. 1
The 21.5% value for Pseudomonas stutzeri RCH2 may reflect strain-specific genomic features rather than a general pattern. 1 These limitations support treating evolutionary loss as a testable interpretation rather than an established fate for all costly, non-conserved genes. 1 The core-gene analysis adds a related limitation: its interpretation of costly-but-conserved genes as targets of purifying selection is based on laboratory fitness and conservation patterns, not direct measurement of natural selection. 2 Fitness Browser condition types are biased toward experimentally convenient conditions rather than ecologically relevant conditions. 2
The new comparison adds that fitness measurements were biased toward rich media and standard stresses, covered 43 bacteria primarily from Proteobacteria, and used single-gene knockouts that did not capture epistatic interactions. 5 Its 16-percentage-point conservation gradient was statistically robust but fitness importance remained only a weak predictor of conservation, and singleton neutrality may reflect poor transposon coverage. 5 These limitations refine the evolutionary-loss interpretation: laboratory burden and conservation are informative complementary signals, not a direct test of natural gene loss. The new linkage analysis adds that the main Escherichia coli clade was absent because it contained too many genomes, Keio mapped to a small clade at only 26.1% coverage, and Dyella79 was excluded because a locus-tag mismatch produced a 0% join rate. 6 These coverage and identifier limitations further support caution when interpreting apparent conservation differences. 6
Open Directions
- Reanalyze gene presence as a continuous fraction of pangenome genomes, rather than a binary core/accessory label, and test whether increasing laboratory burden predicts decreasing prevalence after controlling for organism and gene length. 1
- Reanalyze Fitness Browser measurements with replicate-level models and a continuous burden score rather than the max_fit > 1 rule to test whether MGE-associated genes retain a fitness deficit after reducing single-experiment noise. 1
- Re-link genes using lower-identity and profile-based homology searches, sequence-similarity-sensitive methods, and manual synteny checks to determine how many apparent costly+dispensable, orphan, and singleton assignments were missed or misclassified by the 90% identity DIAMOND threshold. 1
- Longitudinally test whether costly+dispensable genes are lost, retained, or pseudogenized along closely related strain phylogenies, while comparing them with the more prevalent Costly + Conserved trade-off class. 1,2
- Compare long-read assemblies, element-boundary calls, gene neighborhoods, mobile-element annotation, and phylogeny for Pseudomonas stutzeri RCH2 and related strains to distinguish recent invasion from strain-specific genomic expansion. 1
- Measure the 14.1% condition-specific subset across community, host-associated, environmental, and other conditions not represented in the laboratory dataset using targeted competition assays to ask whether context-specific benefits explain persistence. 1
- Integrate MGE annotation, ortholog prevalence, and phylogeny across a larger organism set to test whether the observed burden is general or concentrated in particular lineages. 1
- Stratify the cost–conservation relationship by functional category and condition, especially Cell Wall versus Protein, Motility, and RNA Metabolism genes, using existing Fitness Browser measurements to test whether the +6.2, +7.8, +12.9, and -14.1 percentage-point differences persist after controlling for assay coverage and gene length. 2
- Characterize the 48 accessory fitness modules containing co-regulated functions exclusively in the flexible genome, testing whether their prevalence and laboratory burden predict persistence or loss across related strains. 3
- Reanalyze MGE-associated genes jointly by fitness-effect breadth, conservation class, and condition-specific phenotype to test whether broad or ephemeral fitness effects explain the apparent persistence of burdensome accessory genes. 5
- Test whether the weak fitness–conservation gradient persists after restricting analyses to well-covered genes and adding epistasis-aware, non-rich-media assays, particularly for singleton and condition-specific genes. 5
- Reanalyze the 3,683 essential-auxiliary genes and 1,259 essential-unmapped genes using continuous pangenome prevalence, identifier-resolved orthology, and condition-specific fitness assays to test whether poorly annotated essential genes are retained, replaced, or lost through distinct evolutionary routes. 6
- Extend the essentiality comparison beyond binary calls by correlating mean fitness effects with pangenome prevalence and testing whether genes with fitness < -2 under stress conditions show a different cost–conservation relationship. 6
- Map singleton and auxiliary genes in latent carbon-utilization pathways and in amino acid biosynthesis pathways (24.3% vs 6.5% latent) onto strain phylogenies, and test whether they are real deletion events. Ask whether carbon-pathway loss is concentrated in nutrient-rich lineages, as the evolutionary-lag and Black Queen predictions propose. Use gene-level erosion rather than pathway completeness, because pathway-level conservation did not separate latent from active pathways. 9