Identity
The canonical name is Metal Fitness Atlas. 1
Known aliases are not reported in this document. 1
No stable external identifier is reported in this document. 1
Metal Fitness Atlas is a genome-derived resource that assigns bacterial metal-tolerance scores from annotated metal-tolerance gene clusters. 1 The BacDive phenotype analysis refines this description: its scores are genome-based predictions, and the number of metal-resistance gene clusters (n_metal_clusters) was the strongest predictor in the integrated model. 2
The cross-species fitness atlas broadens the resource beyond composite genome scores: it combines 559 metal experiments across 31 organisms and 16 metals with 383,349 gene × metal fitness records, pangenome conservation, conserved ortholog-family discovery, module analysis, and genome-scale prediction. 3 Its central result is that metal-important genes are predominantly core-genome genes, supporting a core-genome robustness model while distinguishing general metal-sensitive functions from specialized resistance. 3
The metal-specificity analysis refines that distinction: among 7,609 metal-important gene records from 24 organisms, 4,177 (54.9%) were metal-specific, 2,888 (38.0%) were general sick, and 544 (7.2%) were metal+stress. 4 Metal-specific genes were 84.8% core (2,969/3,500), compared with 90.2% (2,183/2,420) for general-sick genes; the difference was significant by Cochran-Mantel-Haenszel testing (p=0.011), supporting a two-tier model in which specialized metal functions are less core-enriched than broad stress functions while both remain core-associated. 4
The metal-cross-resistance analysis supports the relevance of genome-derived metal-response signatures: across 452 metal experiments in 37 organisms, it identified 119,561 genes with metal-fitness data from 28 organisms tested against at least 3 metals. 5 Among 8,162 metal-important genes, 1,484 (18.2%) were general stress genes, 2,306 (28.3%) metal-shared genes, and 4,372 (53.6%) metal-specific genes; their mean pangenome core fractions were 92.0%, 91.0%, and 89.8%, respectively. 5 This supports a three-tier interpretation from broadly conserved stress functions toward more accessory specialized metal responses, but it does not replace the Atlas’s composite score with direct metal-tolerance measurements. 5,2
Cross-species fitness atlas
The new atlas refines the earlier accessory-resistance framing: across 22 organisms and 14 metals, metal-important genes were 87.4% core versus 76.9% for baseline genes (OR=2.08, p=4.3e-162), with 21 of 22 organisms more core-enriched than baseline. 3 This contrasts with the prior condition-specific heavy-metal result of 71.2% core because the genome-wide definition captures cell-envelope functions, DNA repair, central metabolism, protein quality control, and general stress response. 3 Essential-metal tolerance genes had a mean core-fraction delta of +0.148 versus +0.081 for toxic metals (one-sided Mann-Whitney U, U=39, p=0.015), although rare-metal estimates were often based on limited organism coverage. 3
The atlas contained 12,838 broad metal-important genes (3.3%) and 5,667 strict metal-important genes (1.5%) among 383,349 gene × metal records from 24 organisms. 3 Among 2,891 ortholog groups with metal phenotypes, 1,182 were conserved across at least two organisms and 601 across at least three; 149 conserved candidates lacked full functional annotation. 3 Responsive modules were also core-enriched: 183 modules with conservation data had mean core fraction 0.826 and median 0.929. 3
The atlas refines genome-content prediction: a 1,286-KEGG-KO metal signature scored 27,702 pangenome species, but simple gene-presence repertoire scores failed to predict metal fitness. 3 After genome-size normalization, Leptospirillum ranked at the 91st percentile, Acidithiobacillus at the 77th, Marinobacter at the 75th, and Sulfobacillus at the 71st; bioleaching genera were not significantly enriched over background (Mann-Whitney p=0.17). 3 The report proposes a two-tier model in which core general-stress functions dominate broad fitness defects, while accessory efflux, sequestration, and detoxification functions represent specialized resistance. 3
The metal-specificity analysis refines the atlas’s broad categories by showing that specificity depends on metal: the metal-specific fractions were 60.6% for manganese (20/33), 60.5% for molybdenum (185/306), 56.5% for tungsten (173/306), 46.4% for selenium (64/138), 55.9% for cadmium (52/93), 51.9% for copper (1,346/2,594), 50.2% for cobalt (1,167/2,324), 49.3% for chromium (132/268), 48.6% for uranium (88/181), 47.2% for zinc (843/1,786), 43.7% for nickel (993/2,271), 42.4% for aluminum (752/1,772), 32.7% for mercury (35/107), and 21.9% for iron (144/659). 4 Metal-specific genes matched metal-resistance keywords at 12.2% versus 7.8% for general-sick genes (Fisher exact OR=1.64, p=2.4e-8), supporting their biological distinction. 4
The analysis also adds candidate-level prioritization: UCP030820 was metal-specific in 2/3 organisms across 7 metals (67%) with mean sick rate 0.021; YebC in 7/12 organisms across 6 metals (58%) with mean sick rate 0.056; and DUF1043/YhcB in 3/6 organisms across 5 metals (50%) with mean sick rate 0.054. 4 These were prioritized over more pleiotropic candidates including YfdZ, MlaD, YrbC, DUF39, and YrbE. 4 The proposed YebC translation-bottleneck mechanism is explicitly a hypothesis, not an established function. 4 Among 149 novel metal-candidate families, 45.6% were dominantly metal-specific versus 58.2% of annotated families (Fisher exact OR=0.60, p=0.003), indicating that novel families were less metal-specific in this analysis. 4
Fitness-atlas robustness to shared stress
The original Metal Fitness Atlas conclusion that metal-important genes are 87.4% core, with OR=2.08, is supported by a counter-ion analysis: after removing genes shared between metal and NaCl fitness responses, core-genome enrichment remained for 12 of 14 metals. 6 This analysis identified 4,304 of 10,821 metal-important gene records (39.8%) as also NaCl-important, but the corrected atlas signal was preserved rather than explained away by shared NaCl stress. 6
A broader discoveries synthesis refines the interpretation of metal-important records: across 7,609 records, 4,177 (55%) were classified as genuinely metal-specific, 38% as general sick, and 7% as metal-plus-stress. 7 These totals use a different classification analysis from the NaCl-overlap calculation and therefore do not replace its 39.8% estimate. 7
The correction strengthened enrichment for molybdenum (+0.132 to +0.145), tungsten (+0.129 to +0.134), mercury (+0.116 to +0.133), selenium (+0.115 to +0.131), nickel (+0.088 to +0.098), chromium (+0.056 to +0.069), and uranium (+0.031 to +0.040); manganese remained +0.182 and cobalt +0.076. 6 It weakened enrichment for aluminum (+0.099 to +0.068), zinc (+0.145 to +0.115), and copper (+0.090 to +0.084). 6 Cadmium changed from -0.008 to -0.108 and iron from -0.040 to +0.182, but each estimate had low statistical power: cadmium had n=92 genes and iron n=9 genes, each from one organism; manganese also had only n=30 genes from one organism, with its 100% core fraction a ceiling result. 6
The counter-ion analysis therefore refines the Atlas interpretation: researchers do not need to filter out NaCl-responsive genes for the core-enrichment conclusion, although shared-stress biology is substantial and the underlying mechanistic categories remain untested by formal functional-enrichment analyses. 6 The broader analysis similarly found that metal-specific genes were 84.8% core versus 90.2% for general-sick genes, with Cochran–Mantel–Haenszel p=0.011. 7
The metal-cross-resistance study refines this shared-stress interpretation by separating general stress, metal-shared, and metal-specific gene tiers across organisms rather than treating all metal-important genes as one group. 5 It found 318 ortholog groups that were metal-shared in at least 2 metals and at least 2 organisms, with conserved families spanning cell-envelope, energy-metabolism, DNA-repair, and ion-homeostasis functions. 5
Evidence from BacDive validation
The project linked Metal Fitness Atlas scores to bacdive isolation-environment metadata through gtdb species matching. 1
Of 97,334 BacDive strains, 42,227 (43.4%) matched 6,426 GTDB species with metal-tolerance scores, and 25,089 matched strains had isolation-source metadata. 1 A complementary phenotype bridge matched 37,368 strains (38.4%) to 5,647 GTDB species and retained 3,994 species with at least 5 phenotype features. 2 This refines rather than replaces the earlier coverage estimate: the analyses used different phenotype and matching filters. 2
Species-name matching produced 33,535 exact matches (34.5%) and 8,692 additional matches (8.9%) after GTDB suffix removal, while 55,107 strains (56.6%) remained unmatched. 1 The newer study similarly reports that species-name matching recovered 5,647 of 27,702 GTDB species (38.4%) and notes that GCA accession matching was not implemented. 2
Erratum. The report gives 38.4% for two different denominators. 37,368 of 97,334 matched strains is 38.4%; 5,647 of 27,702 GTDB species, the figure this sentence uses, is about one fifth, roughly half what the percentage suggests. 2
Bacteria isolated from heavy-metal contamination sites had a median metal score of 0.240, a mean of 0.236, a median delta of +0.053 relative to the environmental baseline, Cohen’s d of +1.00, and Mann–Whitney p=0.006 (n=10). 1 The environmental baseline had a median metal score of 0.187 and a mean of 0.195. 1 The reported contamination gradient was heavy metal (+1.00) > waste/sludge (+0.57) > all contamination (+0.43) > industrial (+0.20). 1
Waste/sludge isolates had n=305, median 0.219, mean 0.218, median delta +0.032, Cohen’s d=+0.57, and p<0.0001; all-contamination isolates had n=176, median 0.215, mean 0.211, median delta +0.028, Cohen’s d=+0.43, and p<0.0001. 1 Industrial isolates had n=796, median 0.199, mean 0.202, median delta +0.012, Cohen’s d=+0.20, and p<0.0001; host-associated isolates had n=12,086, median 0.194, mean 0.201, median delta +0.007, Cohen’s d=+0.14, and p<0.0001. 1
Within Pseudomonadota, contamination isolates had metal scores delta=+0.040 above environmental isolates (p<0.001; n=85 contamination and n=5,655 environmental), while within Actinomycetota the delta was +0.035 (p<0.001; n=62 and n=772, respectively). 1 The contamination signal was not significant within Bacillota (delta=-0.012, p=0.285; n=19 and n=492) or Bacteroidota (delta=-0.008, p=0.456; n=5 and n=156). 1
The score is genome-size-normalized as metal clusters divided by annotated clusters. 1 For n=10 heavy-metal isolates versus approximately 5,000 environmental-baseline strains, the minimum detectable effect size at 80% power was approximately d=0.93, and the observed d=1.00 barely exceeded this threshold. 1 The report interprets the contamination-associated score pattern as ecological validation of genome-based metal-adaptation predictions, but treats the heavy-metal estimate as imprecise because the matched group contained only n=10 isolates. 1
The phenotype study supports the importance of genome-encoded resistance: taxonomy explained 35.4% of variance, phenotype features alone explained 16.3%, taxonomy plus phenotype yielded R² = 0.345, and adding n_metal_clusters raised the full model to R² = 0.633. 2 In 5-fold phylogenetic-blocked cross-validation, the gene-count-only, phenotype-only, taxonomy-only, taxonomy-plus-phenotype, and full models had R² values of 0.063, 0.163, 0.354, 0.345, and 0.633, respectively, with RMSE values of 0.045, 0.043, 0.038, 0.038, and 0.028. 2
The phenotype results refine interpretation of the Atlas as a genome-derived score: Gram stain, oxidase, motility, urease, enzyme breadth, nitrate reduction, and catalase were significant after FDR correction, but SHAP (Shapley additive explanation) importance placed taxonomic codes and n_metal_clusters above phenotype features, indicating that many phenotype associations were phylogenetic proxies rather than independent predictors. 2 The Gram-negative association was strong overall (Cohen’s d = -0.61, p < 1e-60, n = 3,272 species) but was largely between-lineage, while the urease association reversed the predicted direction and disappeared within Gammaproteobacteria and Bacilli. 2
A separate BacDive analysis supports the contamination association but adds a confounding caution: heavy-metal-contaminated isolates had d=+1.00, p=0.006, n=10, whereas host-associated bacteria unexpectedly scored higher than environmental bacteria at d=+0.14, p<0.0001, likely because host-associated samples were enriched for large-genome Pseudomonadota pathogens. 7
The new cross-resistance analysis contradicts a strong version of ecological validation at the small organism-matched scale: multi-metal tolerance scores showed Spearman rho approximately -0.02 with BacDive metal-associated isolation at the Fitness Browser species scale (p > 0.8), and the matched analysis retained only 20 independent species after exclusions and strain collapsing. 5 This does not contradict the pangenome-scale Atlas result of Cohen’s d = +1.0 from 42K strains, but refines the evidence boundary: the organism-scale null is underpowered and is not directly comparable to the 42K-strain validation. 5
Limitations and planned extensions
Only 24 matched strains had metal-utilization records covering iron, manganese, arsenate, chromate, cobalt, or zinc; 8 results were positive and 16 negative, and the exploratory comparison was inconclusive (Mann–Whitney p=0.14; Cohen’s d=-0.57). 1
The report identifies direct BacDive GCA-accession matching to pangenome genome IDs as a way to recover more of the 56.6% unmatched strains. 1 The phenotype study likewise recommends GCA accession matching and phylogenetic generalized least squares (PGLS) or phylogenetic PCA to separate phenotype effects from lineage structure. 2 It also proposes integrating ENIGMA CORAL community data, testing metal-tolerance gene families against specific contamination environments, and expanding phenotype extraction to MIC and growth-inhibition data. 1
The Atlas scores remain genome-based predictions rather than direct metal-tolerance measurements; the phenotype analysis therefore establishes phenotype-to-genome correlations, not phenotype-to-phenotype validation. 2 Its composite score cannot establish metal-specific mechanisms, so per-metal scores and matched experiments are needed to test whether catalase predicts copper, urease predicts nickel, or H₂S predicts zinc, copper, and cadmium tolerance. 2 The direct validation contained 12 organisms representing 6 unique species and was underpowered for within-set Gram comparisons. 2
The new atlas adds several limitations: metal concentrations were not normalized by dose or tolerance threshold; uranium, chromium, mercury, cadmium, selenium, and manganese had only one or two organisms of coverage; and phylogenetic non-independence may inflate cross-species conservation, although excluding four duplicate P. fluorescens FW300 strains retained the core-enrichment result (OR=2.065, p=5.9e-141). 3 Its broad definition, fit < -1 OR n_sick ≥ 1, includes general-stress genes, and putatively essential genes—approximately 14.3% of protein-coding genes and approximately 82% core—lack transposon insertions, making the observed enrichment a conservative estimate. 3
The metal-specificity analysis adds 40.7% gene attrition because locusId mismatches excluded ANA3, Dino, Keio, MR1, Miya, PV4, and SB2B; the 24 included organisms represented 7,609 of 12,838 records (59.3%), so excluded organisms may have different specificity profiles. 4 The 5% sick-rate threshold was arbitrary but results were qualitatively stable across 1–20%; validation against Fitness Browser specificphenotype annotations was not performed. 4 Its independent component analysis (ICA), a method for decomposing activity profiles into modules, identified 0 metal-specific modules because raw module-condition scores were on a different scale from the precomputed z-scored profiles used in the Metal Atlas analysis. 4 Proposed follow-up includes resolving locusId mismatches, repeating the counter-ion analysis with fit < -1 without the |t| > 4 requirement, using precomputed z-scored module activities, and applying AlphaFold predictions to UCP030820, YebC, and DUF1043. 4
The counter-ion robustness analysis was limited by NaCl threshold dependence, incomplete cross-organism replication for manganese, cadmium, selenium, mercury, iron, molybdenum, and tungsten, and the fact that NaCl supplies both Na⁺ and Cl⁻ and causes osmotic stress. 6 It recommends formal COG, KEGG, and PFAM enrichment tests, choline-chloride comparisons, independent component analysis (ICA), and matched CuCl₂/CuSO₄, ZnCl₂/ZnSO₄, and CoCl₂/CoSO₄ RB-TnSeq experiments under identical conditions. 6
The discoveries synthesis refines the need for mechanistic validation: zinc sulfate, which contains 0 mM chloride, had 44.6% overlap with NaCl-important genes and the highest DvH profile correlation, r=0.715; excluding SynE, an experiment-count outlier with 88.6% overlap, reduced the overall overlap rate to 36.7%. 7
The cross-resistance study further cautions that metal concentrations differed among experiments, organisms contributed between 3 and 112 metal experiments, and the 28 organisms were not phylogenetically independent. 5 It recommends pangenome-scale validation across 27K species, PGLS or independent contrasts, normalization by metal concentration relative to MIC, ICA of metal-condition modules, and AlphaFold-based structural analysis of metal-shared proteins. 5
The atlas likewise recommends regulatory or expression-based models, concentration-relative-to-MIC normalization, phylogenetic independent contrasts, enrichment-based pangenome scoring, and functional annotation of the 149 novel candidates using PaperBLAST, InterPro, and structural prediction. 3
Related pages
- BacDive Isolation Environment × Metal Tolerance Prediction — source report for the BacDive validation analysis. 1
- BacDive Phenotype Signatures of Metal Tolerance — phenotype-to-metal-score integration and phylogenetic-confounding analysis. 2
- Counter Ion Effects on Metal Fitness Measurements — analysis of shared NaCl stress, counter-ion confounding, and Atlas core-enrichment robustness. 6
- discoveries — cross-project synthesis of metal-specificity, shared-stress, and BacDive validation findings. 7
- Gene-Resolution Metal Cross-Resistance Across Diverse Bacteria — gene-resolution analysis of conserved cross-resistance, gene tiers, and organism-scale BacDive validation. 5
- Pan-Bacterial Metal Fitness Atlas — cross-species metal-fitness experiments, conserved gene families, responsive modules, and pangenome prediction. 3
- Metal-Specific vs General Stress Genes — classification of metal-specific versus general-stress genes, candidate prioritization, and module-analysis limitations. 4
- environmental-resistome — ecological validation and genome-encoded predictors of metal tolerance. 1
- pangenome-integration — GTDB pangenome matching coverage and unmatched-strain limitations. 1
- multi-omics-integration — integration of genome-derived scores with curated isolation and utilization phenotypes. 1
- bacdive — source of strain isolation and metal-utilization metadata.
- gtdb — species framework used for matching BacDive strains to pangenome scores.