Separating Shared Stress Responses from Stressor-Specific Fitness Requirements

Fitness profiles measured under different stressors can contain both a shared cellular-stress component and requirements specific to the tested stressor. This distinction matters when interpreting condition-specific fitness. Overlap between stressor responses does not by itself show that the delivery vehicle or counter ion caused the signal. Likewise, a salt-associated fitness signal should not automatically be read as a specific response to the intended metal. ₁ This framework refines condition-specific-fitness by requiring metal fitness profiles to be decomposed into shared-stress and metal-specific components before mechanistic interpretation. ₁

Evidence for a Shared-Stress Component

The Counter Ion Effects on Metal Fitness Measurements directly compares metal and NaCl fitness data. The comparison covered 71 NaCl experiments, 10,821 metal-important gene records, 86 organism × metal pairs, 14 metals, and 19 organisms. Of those records, 4,304 (39.8%) were also NaCl-important. ₁ The analysis therefore supports a substantial shared-stress component, while retaining a larger metal-specific component of 6,517 records (60.2%). ₁ This pooled gene-record percentage is distinct from the per-metal, organism-level means reported below. ₁

NaCl-important genes overlapped with metal-important genes for every metal, with overlap ranging from 9.2% for molybdenum to 57.6% for manganese. ₁ The source table labels these per-metal percentages as mean overlap and gives a separate number of organisms for each metal. They must not be conflated with the pooled 4,304/10,821 gene-record overlap. ₁ The complete per-metal values were as follows. ₁

• manganese 57.6% (1 organism)
• cadmium 50.5% (1)
• aluminum 46.1% (12)
• zinc 44.6% (12)
• cobalt 41.3% (18)
• copper 41.0% (16)
• nickel 39.3% (17)
• uranium 37.2% (2)
• iron 33.3% (1)
• chromium 29.2% (2)
• selenium 26.8% (1)
• mercury 23.4% (1)
• tungsten 10.8% (1)
• molybdenum 9.2% (1) ₁

The same table pairs organism-level mean overlap with mean maximum Cl⁻ concentration. Zinc had 44.6% across 12 organisms at 0 mM, cobalt 41.3% across 18 at 28 mM, copper 41.0% across 16 at 2 mM, and nickel 39.3% across 17 at 2 mM. ₁

Removing the Synechococcus elongatus outlier reduced overlap from 39.8% to 36.7%, or from 4,304/10,821 to 3,739/10,183 records, so the overall signal was not driven by that organism. ₁ The estimate is still especially sensitive to S. elongatus (SynE). SynE contributed 565/638 shared genes (88.6%) from 12 NaCl dose-response experiments spanning 0.5–250 mM, far more than any other organism (1–6 NaCl experiments). The n_sick ≥ 1 rule is much easier to satisfy with 12 experiments. _1,2 The discoveries digest therefore advises that analyses using NaCl fitness data account for experiment count when setting importance thresholds. This coverage effect is discussed in fitness-condition-coverage-prioritization-bias. ₂ The pitfalls digest generalizes the problem. The n_sick >= 1 rule is biased by the number of NaCl experiments per organism. SynE, with 12 NaCl dose-response experiments (0.5–250 mM), flags 32.6% of genes as NaCl-important, 3× higher than the next organism, and shows 88.6% metal–NaCl shared-stress. ₃ It recommends three remedies: requiring n_sick >= 2, using a mean_fit threshold, or reporting results with and without outlier organisms. Excluding SynE lowers overall metal–NaCl overlap from 39.8% to 36.7%, which it calls modest but worth documenting. ₃

Shared-stress genes were important for a mean of 4.1 metals per gene, compared with 2.5 metals per metal-specific gene. ₁ In Desulfovibrio vulgaris Hildenborough (DvH), 495 unique metal-important genes comprised 73 shared-stress genes (14.7%) and 422 metal-specific genes (85.3%). ₁ This supports the interpretation that shared-stress genes tend to recur across more metal conditions, whereas metal-specific genes contribute more narrowly distributed fitness requirements. ₁

The Gene-Resolution Metal Cross-Resistance Across Diverse Bacteria supports this shared component from a different angle. It examined 317 organism–metal-pair observations involving 28 organisms and 85 unique metal pairs. Of these gene-level fitness correlations, 98.1% were positive (311/317), and 99.1% were statistically significant (p < 0.05). ₄ The report’s limitations section states that all tested metal pairs are positive. That statement conflicts with the report’s own 311/317 count, and both source statements are preserved here. ₄ The metal-label permutation test was non-significant (p = 0.42), and the study had no non-metal stress controls. It therefore cannot distinguish universal metal cross-resistance from general stress shared by all metals, a gap the report says the counter-ion project only partially addresses. This evidence refines rather than independently identifies a shared-stress mechanism. ₄

The Pan-Bacterial Metal Fitness Atlas supports this decomposition at atlas scale. The atlas inventory spans 559 experiments, 31 organisms, and 16 metals. Its core-genome analysis covered 22 organisms and 14 metals. In that analysis, metal-important genes were predominantly core-genome genes: 87.4% core versus 76.9% for baseline genes (OR=2.08, p=4.3e-162). ₅ The atlas proposes the two-tier model outlined below. ₅

• Tier 1 (core) is the general stress response. These genes are the majority of metal fitness genes. They are conserved because they serve essential cellular functions beyond metal tolerance.
• Tier 2 (accessory) is specific metal resistance. This smaller fraction provides targeted tolerance via efflux, sequestration, or enzymatic detoxification. These genes are enriched in the accessory genome and are visible only when controlling for general stress response. ₅

This refines the overlap analysis by showing that shared stress does not imply that the entire metal fitness signal is nonspecific. ₅ The atlas’s own conservation analysis did not, however, isolate metal-specific genes from genes important under other stresses. Its first proposed follow-up is to repeat the conservation analysis using only genes important for metals but not for other stresses, to separate the core stress-response signal from genuine metal-resistance gene evolution. ₅

The functional interpretation is consistent with shared cellular pressures involving cell-envelope integrity, DNA repair, and ion homeostasis. These categories, however, were proposed from fitness profiles rather than tested with formal functional-enrichment analyses or directly tested biochemically. ₁ These mechanistic assignments should therefore be treated as hypotheses rather than established pathway findings. ₁

The Metal-Specific vs General Stress Genes supports this decomposition with an independent classification of 7,609 metal-important gene records with fitness-matrix data across 24 organisms. Of these, 4,177 (54.9%) were metal-specific: they showed significant fitness defects under metal stress but a <5% sick rate across 5,945 non-metal experiments. The remainder split into general sick (2,888, 38.0%) and metal+stress (544, 7.2%). ₆ Its metal-specific category used a stricter non-metal sick-rate criterion, so it is not a replacement estimate for the 39.8% NaCl-overlap fraction. ₆

The metal-specific fraction depends on the sick-rate cutoff. About ~41% of records were metal-specific at a 2% cutoff and ~67% at 10%. ₆ The report calls the classification robust across thresholds. Its limitations section nonetheless states that the 5% cutoff is arbitrary, and that results are qualitatively stable across 1-20% while exact fractions vary. ₆ The 54.9% figure should therefore be read as threshold-conditional rather than as a fixed property of the metal-important set. ₆

The two projects’ osmotic-overlap estimates also differ. The counter-ion analysis found 39.8% overlap between metal-important and NaCl-stress genes. The metal-specificity analysis found 14.7% of metal-important genes sick under osmotic stress, a 2.7x discrepancy. ₆ The report attributes the difference to method. It used a stricter threshold (fit < -1 AND |t| > 4) than the counter-ion threshold (fit < -1 only), and the organism sets partly differ. ₆ This refines the shared-stress estimate by showing that the fraction depends materially on thresholds and condition scope. The report proposes validating the osmotic overlap by replicating the counter-ion methodology (fit < -1 without |t| > 4). That validation has not been done, so the methodological explanation remains untested. ₆

The central discoveries digest reports the same classification across 24 organisms. Of 7,609 metal-important gene records, 4,177 (55%) were sick under metal stress but had a <5% sick rate across 5,945 non-metal experiments (antibiotics, osmotic, carbon sources, etc.). The remaining 38% were “general sick” (important across many conditions), and 7% were “metal+stress” (shared with other stresses). ₂ The digest takes this to show that general stress response does not dominate the Metal Fitness Atlas’s metal-important gene set. In its reading, the majority are genuine metal tolerance determinants. That interpretation is in tension with the atlas’s own limitations wording (see Tensions). _2,5

Evidence for Stressor-Specific Requirements

The overlap did not account for most metal-important records: 6,517 of 10,821 records (60.2%) were classified as metal-specific, meaning important for metals but not NaCl. ₁ In DvH, 495 unique metal-important genes split into 73 shared-stress (14.7%) and 422 metal-specific (85.3%) genes. Shared-stress genes were important for a mean of 4.1 metals, versus 2.5 for metal-specific genes. ₁ Also in DvH, 90.5% of metal-specific genes had SEED annotations, compared with 78.1% of shared-stress genes. The report suggests that the metal-specific set includes well-characterized metal homeostasis functions (Ni/Fe-hydrogenase, nitrogenase regulators, metal transporters). This comparison was descriptive rather than a formal enrichment test. ₁

The metal-specificity analysis supports a stressor-specific interpretation. Metal-specific genes were 1.64x more likely than general-sick genes to match metal-resistance keywords such as efflux, transporter, metal, CDF, and siderophore (12.2% versus 7.8%; Fisher exact OR=1.64, p=2.4e-8). The functional-enrichment table’s annotated denominators were 3,344 metal-specific and 2,573 general-sick genes. ₆

The same table complicates the general-stress side of the split. Among 3,344 annotated metal-specific genes, 12.2% matched metal-resistance keywords and 13.7% matched general-stress keywords (DNA repair, cell wall, chaperone, etc.). Among 2,573 annotated general-sick genes, 7.8% matched metal-resistance keywords and 11.5% matched general-stress keywords. Among 495 annotated metal+stress genes, 8.9% matched metal-resistance keywords and 6.5% matched general-stress keywords. ₆ The report’s prose nonetheless states that general-sick genes show slightly higher general-stress keyword enrichment “at 11.5% vs 13.7%”, which its own table values contradict, since the table assigns 13.7% to metal-specific and 11.5% to general-sick genes. ₆ Keyword matching therefore supports only the metal-resistance side of the expected split. It does not show general-stress functions concentrated in general-sick genes, and the internal inconsistency is preserved here rather than resolved. ₆

The metal-specificity analysis also reports specificity for each metal, with DvH included after its locusId fix. The list below gives metal-specific records over metal-important records. ₆

• manganese (essential): 20/33, 60.6%
• molybdenum (essential): 185/306, 60.5%
• tungsten (essential): 173/306, 56.5%
• cadmium (toxic): 52/93, 55.9%
• copper (toxic): 1,346/2,594, 51.9%
• cobalt (toxic): 1,167/2,324, 50.2%
• chromium (toxic): 132/268, 49.3%
• uranium (toxic): 88/181, 48.6%
• zinc (toxic): 843/1,786, 47.2%
• selenium (essential): 64/138, 46.4%
• nickel (toxic): 993/2,271, 43.7%
• aluminum (toxic): 752/1,772, 42.4%
• mercury (toxic): 35/107, 32.7%
• iron (essential): 144/659, 21.9% ₆

The report concludes that essential metals show substantial specificity, citing manganese (60.6%), molybdenum (60.5%), tungsten (56.5%), and selenium (46.4%). ₆ Its prose also states that toxic metals range from 42-56% metal-specific, which conflicts with its own mercury entry of 35/107 (32.7%). It places iron lowest at 21.9% and proposes, without testing, that this reflects iron’s central role in core metabolism. ₆ These table values refine the rounded essential-metal specificities in the discoveries digest discussed below. _6,2

DvH, the most extensively profiled organism, supplied six NaCl experiments for comparison with profiles for 13 metals. ₁ Its whole-genome Pearson correlations with NaCl formed a heterogeneous hierarchy rather than a uniform shared-stress response. The values were zinc r=0.715, manganese r=0.545, copper r=0.532, cobalt r=0.498, mercury r=0.478, nickel r=0.446, aluminum r=0.420, molybdenum r=0.396, uranium r=0.350, selenium r=0.342, chromium r=0.318, tungsten r=0.298, and iron r=0.086. ₁ Zinc ranked first despite being delivered as sulfate with zero Cl⁻, so the hierarchy does not follow Cl⁻ concentration. ₁ Higher correlations for zinc, manganese, copper, cobalt, mercury, and nickel are consistent with broad toxicity through essential-cofactor displacement or disruption of multiple cellular systems. Lower correlations for molybdenum, uranium, selenium, chromium, tungsten, and iron are consistent with more pathway-specific effects. ₁ These mechanistic interpretations are extrapolations from fitness profiles, not direct biochemical findings, and remain hypotheses. ₁ The discoveries digest reports the same hierarchy with rounded values, from Zn (r=0.72) down to Fe (r=0.09). It interprets the hierarchy as a general-to-specific toxicity gradient. ₂ The DvH result supports examining shared-condition architecture in cofitness-network-architecture. However, a whole-genome correlation cannot identify whether shared genes respond to chloride, sodium, osmolarity, the metal, or interactions among these stresses. ₁

The metal cross-resistance report states that the counter-ion project found DvH’s metal–NaCl hierarchy followed toxicity mechanism rather than counter-ion identity. It interprets its own metal–metal correlations across 28 organisms as showing a similar mechanistic grouping. ₄ This supports the toxicity-mechanism reading, but both studies make the same interpretive step from fitness correlations, and neither tested mechanism biochemically. _4,1

The metal cross-resistance study refines the two-tier model by classifying 8,162 metal-important genes across 28 organisms into the three tiers listed below. ₄

• General stress: 1,484 genes (18.2%), mean 92.0% core, 57.2% fully core (≥95%)
• Metal-shared, important for ≥2 metals: 2,306 genes (28.3%), mean 91.0% core, 50.4% fully core (≥95%)
• Metal-specific, important for exactly 1 metal: 4,372 genes (53.6%), mean 89.8% core, 45.7% fully core (≥95%) ₄

Pangenome conservation therefore declined from general-stress to metal-shared to metal-specific genes. The fully-core proportions of 57.2%, 50.4%, and 45.7% span 11.5 percentage points. ₄ The report presents this as an extension of the metal-fitness-atlas estimate that metal genes are 87.4% core. Within metal genes it finds a finer gradient: general stress (92%) > shared (91%) > specific (90%). ₄ This supports a gradient from broadly conserved stress functions toward more specialized requirements. It also shows that even stressor-specific fitness requirements can remain highly conserved. ₄

Functional keyword analysis in the same study associated general-stress genes with energy/respiration and cell-envelope functions. It associated metal-specific genes with transporters/efflux and iron/metal-related functions. ₄ This partially supports the envelope-versus-efflux split that the counter-ion analysis proposes but did not formally test with annotation-enrichment statistics. A keyword analysis is not a substitute for that test. _4,1

The discoveries digest partially supports the atlas’s two-tier model from the metal-specificity classification. Metal-specific genes were 84.8% core (pooled) versus 90.2% for general-sick genes, and both categories exceed the 79.8% baseline. The difference was significant under the Cochran–Mantel–Haenszel test, a stratified test of association (p=0.011). ₂ General stress response (Tier 1) is thus more core than specialized metal resistance (Tier 2), but both are predominantly core. The digest concludes that the accessory genome contributes only modestly to metal resistance machinery. ₂ The counter-ion report likewise attributes to metal_fitness_atlas the finding that genes important only under metal conditions are less core. It proposes that its own shared-stress classification could refine that analysis. ₁

The metal-specificity report’s conservation table supports the digest’s account and adds per-category detail. Metal-specific genes were 84.8% core pooled (2,969/3,500) and 88.0% by organism mean, with a +6.9% mean delta. For this category, 19/22 organisms had positive deltas and 12/22 were significant. ₆ General-sick genes were 90.2% core pooled (2,183/2,420) and by organism mean, with a +9.0% mean delta. For this category, 21/21 organisms had positive deltas and 8/21 were significant. ₆ Metal+stress genes were 94.3% core pooled (467/495) and 93.6% by organism mean, with a +10.9% mean delta. For this category, 13/13 organisms had positive deltas but only 1/13 was significant. ₆ A Cochran–Mantel–Haenszel test found metal-specific genes less core-enriched than general-sick genes across organisms (p=0.011). ₆ This refines the two-tier model. The most metal-specific category is the least core, while the small metal+stress category is the most core yet rarely significant in individual organisms. Per-organism support is therefore uneven across categories. ₆

The atlas supports the existence of stressor-specific requirements but refines their interpretation. Essential-metal tolerance genes for Fe, Mo, W, Se, and Mn had a mean core-fraction delta of +0.148, compared with +0.081 for toxic metals (Mann-Whitney U=39, p=0.015). ₅ Thus, stressor-specific fitness requirements need not be accessory resistance genes; many can be conserved functions whose vulnerability differs among metals. ₅ The atlas identified 149 conserved metal-biology candidates without full functional annotation, but these remain function predictions based solely on cross-species fitness data. ₅

Among the atlas’s top novel candidates, the discoveries digest identifies three as metal-specific rather than pleiotropic. The first is ucp030820/OG01015, an oxidoreductase that is 67% metal-specific across 7 metals. The second is yebc/OG01383, a transcriptional regulator/translation factor that is 58% metal-specific across 6 metals and 11 organisms. The third is duf1043 yhcb/OG03264, a cell division protein that is 50% metal-specific across 5 metals. By contrast, YfdZ (15%) and the Mla/Yrb system (0-25%) are pleiotropic, important for many conditions rather than just metals. DUF39 (0%, sick rate 0.64) is a general fitness factor despite spanning 8 metals. ₂ This refines the 149-candidate list by showing that breadth across metals does not by itself indicate metal specificity. _2,5

• ucp030820 (OG01015; 3 organisms, 7 metals): 2/3 records metal-specific (67%), mean sick rate 0.021
• yebc (OG01383; 11 organisms, 6 metals): 7/12 records metal-specific (58%), mean sick rate 0.056
• duf1043 yhcb (OG03264; 6 organisms, 5 metals): 3/6 records metal-specific (50%), mean sick rate 0.054
• UPF0042/RapZ (OG02094; 8 organisms, 7 metals): 2/8 records metal-specific (25%), mean sick rate 0.130
• MlaD (OG04003; 4 organisms, 4 metals): 1/4 records metal-specific (25%), mean sick rate 0.113
• YfdZ (OG00391; 7 organisms, 9 metals): 2/13 records metal-specific (15%), mean sick rate 0.268
• YrbC (OG02233; 8 organisms, 4 metals): 1/9 records metal-specific (11%), mean sick rate 0.234
• DUF39 (OG08209; 2 organisms, 8 metals): 0/2 records metal-specific (0%), mean sick rate 0.637
• YrbE (OG03534; 6 organisms, 5 metals): 0/6 records metal-specific (0%), mean sick rate 0.190 ₆

The list above gives the metal-specificity report’s underlying candidate counts. The report describes DUF1043/YhcB as a cell-division and envelope-coordination protein. ₆ It also proposes a hypothesis, not tested in the analysis, that YebC is specifically required under metal stress. On this hypothesis, metal-induced demand for proline-rich metal transporters creates a translation bottleneck that YebC resolves. ₆ YfdZ and the Mla/Yrb system (YrbC/D/E) were pleiotropic, sick under many non-metal conditions. The report links YfdZ’s sick rate of 0.268 to its known role in alanine biosynthesis. It reads the Mla system’s pleiotropic defects as consistent with outer-membrane maintenance under diverse stresses (see outer-membrane-lipid-homeostasis). ₆ DUF39 showed 0% metal specificity with a sick rate of 0.637, so it is important for many conditions rather than for metals alone. ₆ These classifications rest on fitness sick rates rather than functional tests, and the mechanistic readings remain interpretive. ₆

Analogous Decomposition Beyond Metals

The same shared-versus-specific problem appears in carbon-source fitness data. In the FW300 metabolic-consistency analysis, 231 genes were significant in 3+ metabolite conditions. The top 18 genes, significant in all 21 conditions, were amino acid biosynthesis enzymes. Examples are homoserine O-acetyltransferase (methionine), ATP phosphoribosyltransferase (histidine), dihydroxy-acid dehydratase (branched-chain amino acids), isopropylmalate dehydrogenase (leucine), shikimate dehydrogenase (aromatic amino acids), and imidazoleglycerol-phosphate dehydratase (histidine). _7,2 These genes are essential for growth on any minimal medium and represent “housekeeping fitness”. By contrast, the ~370 genes significant in only 1-2 conditions were interpreted as the truly substrate-specific ones. Examples include carnitine-specific transporters, lactate dehydrogenases, and trehalose-specific phosphotransferases. _7,2 The fitness signal therefore mixes housekeeping and substrate-specific components. The report recommends separating them before integrating Web of Microbes and Fitness Browser data. By analogy with the metal case, this supports decomposing condition fitness before interpreting it. _7,2

A related caveat comes from the field-versus-lab comparison. Field and lab fitness effects were correlated at roughly r ~ 0.7, read from a scatter plot, so most genes that are sick in one context are sick in the other. This reduces the apparent context specificity of fitness effects. ₈ The value is a visual estimate rather than a reported test statistic. It nonetheless supports the general point that a large shared component can underlie nominally context-specific fitness, as discussed in lab-field-fitness-concordance. ₈

Field metagenomic association data pose the same decomposition problem without fitness assays. The soil metal functional-genomics project ran a copper-specific analysis on 7,566 samples with nearby KBase genomes at 10 km. In it, 116 COGs (Clusters of Orthologous Groups) passed FDR < 0.05, where FDR is the false discovery rate. ₉ Top positive associations included cell-division and nucleotide-transport COGs, and top negatives included energy-production COGs. ₉ The report reads this pattern as suggesting, not establishing, energetic trade-offs under copper stress. ₉

All of those associations are observational. Cr, Cu, Pb, and Zn co-vary in many industrial soils. It is therefore unresolved whether individual COG–metal associations are metal-specific or reflect a generic stress response to multi-metal contamination. ₉ The report planned a partial-correlation model, COG ~ Cr | Cu + Zn + Pb, to separate metal-specific from general responses, but that model is pending. ₉ This makes the shared-versus-specific question newly testable in field data, but the field data currently provide no direct evidence for either side. ₉

Salt Chemistry and the Counter-Ion Test

The shared signal was not explained by chloride dose. Zinc sulfate delivered 0 mM chloride yet had 44.6% NaCl overlap. That exceeds cobalt at 41.3%, which delivered up to 500 mM chloride, as well as copper at 41.0% and nickel at 39.3%. ₁ Chloride-delivered metals had a mean overlap of 41.6%, compared with 37.8% for non-chloride metals. ₁ Among the 14 metals, zinc (sulfate, zero chloride) ranked 4th in NaCl overlap and uranium (acetate, zero chloride) ranked 8th. The report’s reading that the ranking follows toxicity mechanism rather than counter-ion identity is interpretive. ₁ The report therefore rejected H1b, the prediction that overlap scales with Cl⁻ concentration, because ZnSO₄ (0 Cl⁻) exceeded most chloride metals. ₁ In DvH, zinc sulfate also had the highest NaCl correlation, r=0.715, despite delivering zero chloride. ₁ The discoveries digest treats this zinc result (44.6% overlap and r=0.715 at 0 mM Cl⁻) as the definitive evidence against chloride counter ions as the primary confound. It sets this result against the 39.8% metal–NaCl overlap across 19 organisms and 14 metals. ₂

These results support the conclusion that shared fitness signals primarily reflect common cellular stress biology rather than chloride contamination. They do not show that counter ions are universally negligible. ₁ The report rated H0, that counter ions are negligible, as only partially supported. In its assessment, counter ions specifically are negligible, while shared stress biology is real but does not undermine atlas conclusions. ₁ sodium chloride is not a pure chloride control because it supplies both Na⁺ and Cl⁻ and produces osmotic stress. Its fitness profile therefore includes sodium toxicity and osmotic effects beyond chloride. ₁ A KCl or choline chloride control would isolate chloride more specifically. However, the choline chloride experiments available for some Fitness Browser organisms were run at different concentrations. ₁ The evidence therefore argues against chloride as the primary explanation while leaving broader salt, sodium, osmotic, and condition-specific effects unresolved. ₁

As literature context, the report cites Danilova et al. (2020), who reported 2.5-fold differences between CuSO₄ and CuCl₂ toxicity at 0.5 M, a concentration range where osmotic effects dominate. ₁ RB-TnSeq (random barcode transposon sequencing) typically uses lower concentrations, 0.05–2 mM. The report infers that at these concentrations counter-ion effects would be proportionally smaller. That is an extrapolation across concentration ranges, not a measured counter-ion effect. ₁

The only within-organism comparison of copper salts was in psRCH2, and it was severely confounded: CuCl₂ was tested anaerobically (3 replicates) and CuSO₄ aerobically (3 replicates). ₁ The cross-salt correlation was r=0.439, compared with within-replicate correlations of r=0.720 for CuCl₂ and r=0.859 for CuSO₄. ₁ CuSO₄ had a higher NaCl correlation than CuCl₂, r=0.450 versus r=0.212, despite supplying no chloride. ₁ This argues against chloride as the primary confound. However, the aerobic/anaerobic difference could independently affect hundreds of genes, so the comparison cannot isolate counter-ion effects. ₁

Consequences for Atlas-Level Inference

Removing shared-stress genes preserved Metal Fitness Atlas core-genome enrichment for 12 of 14 metals. The report therefore rejected H1c, the prediction that core enrichment would weaken after shared-stress correction. ₁ Corrected enrichment strengthened for the metals listed below. ₁

• molybdenum, from +0.132 to +0.145
• tungsten, from +0.129 to +0.134
• mercury, from +0.116 to +0.133
• selenium, from +0.115 to +0.131
• nickel, from +0.088 to +0.098
• chromium, from +0.056 to +0.069
• uranium, from +0.031 to +0.040 ₁

The discoveries digest repeats the main points. After removing the ~40% shared-stress component, enrichment persists for 12 of 14 metals, with the Mo delta changing from +0.132 to +0.145. The original atlas conclusion (OR=2.08) is not an artifact. ₂

Aluminum weakened from +0.099 to +0.068, zinc from +0.145 to +0.115, and copper from +0.090 to +0.084. Manganese remained +0.182, and cobalt remained +0.076. ₁ Cadmium changed from -0.008 to -0.108, and iron changed from -0.040 to +0.182. ₁ Cadmium had n=92 genes and iron had n=9 genes, each from one organism. Manganese had n=30 genes from one organism, and its 100% core fraction is a ceiling result. ₁ The report’s narrative says core enrichment was preserved for 12 of 14 metals. It describes stronger enrichment for molybdenum (+0.132 → +0.145), tungsten (+0.129 → +0.134), mercury (+0.116 → +0.133), and selenium (+0.115 → +0.131). It describes modest decreases for aluminum (+0.099 → +0.068) and zinc (+0.145 → +0.115). The narrative also says that only cadmium “reverses” (-0.008 → -0.108), which it attributes to very low statistical power given 92 original genes from 1 organism. That wording conflicts with the report’s corrected-conservation table, in which cadmium remains negative and iron is the metal whose delta changes sign. The narrative’s reversal claim should be read with this internal inconsistency in mind. ₁

The original conclusion that metal-important genes were 87.4% core, with OR=2.08, was not attributable to shared NaCl-stress genes. ₁ This supports gene-essentiality: removing a broad shared-stress component does not eliminate the atlas-level conservation signal. Approximately 14.3% of protein-coding genes were classified as putative essential genes because they lacked transposon insertions and were absent from both NaCl and metal-fitness data. These genes were 82% core, and their exclusion affected the original and corrected conservation analyses equally. ₁

The atlas also found that a simple gene-presence repertoire score failed to predict metal fitness. This supports the conclusion that metal importance cannot be inferred from repertoire alone and likely requires regulatory or expression-based models. ₅

Scope, Tensions, and Caveats

The 39.8% overlap depends on the NaCl-importance threshold of fit < -1 or n_sick ≥ 1. A stricter threshold would reduce overlap, and a more permissive one would increase it. ₁ NaCl experiments identified 4,648 NaCl-important genes across 25 organisms, whereas the overlap analysis included 19 organisms and 86 organism × metal pairs. ₁ Manganese, cadmium, selenium, mercury, iron, molybdenum, and tungsten were each tested in only 1 organism (DvH), so their overlap statistics have no cross-organism replication. ₁

The atlas’s own limitations section defines metal-important genes as fit < -1 OR n_sick ≥ 1. This broad rule captures ~3.3% of genes, most of which the atlas describes as general stress genes rather than metal-specific. The definition differs from the fit < -1, |t| > 4 wording used earlier in the atlas. ₅ Definitions should therefore be checked before comparing shared-stress fractions across projects, a schema-dependence discussed in ontology-and-category-schema-sensitivity. ₅

The cross-resistance study likewise cautions about several limits. Metal concentrations differed among experiments, organisms ranged from 3 to 112 metal experiments, and the 28 organisms were not phylogenetically independent. Formal PGLS (phylogenetic generalized least squares) or independent contrasts would strengthen the conservation claim. ₄ Its leave-one-out consensus prediction had mean r = 0.41, but only 2/28 organisms reached individual significance. The conserved chemistry-specific magnitude layer therefore remains less secure than the universal positive direction. ₄

The stricter metal-specificity analysis also had substantial representation limits. Of 12,838 metal-important gene records, 7,609 were retained and 5,229 were excluded because of locusId-format mismatches. Seven metal-tested organisms could not be processed. ₆ This refines interpretation of cross-condition fractions by showing that both classification thresholds and analyzable organism sets can shape the apparent balance between shared and stressor-specific requirements. ₆ The planned validation against the Fitness Browser’s built-in specificphenotype condition-specificity annotations was not performed and remains future work. The classification therefore lacks that independent check. ₆

The discoveries digest records that an initial analysis showed 0% metal-specificity for the essential metals Mo, W, Se, and Mn. This at first seemed to follow from DvH’s 608 non-metal experiments making specificity impossible. The actual cause was a locusId type mismatch that had silently excluded DvH. After the fix, the listed specificities were manganese (61%), molybdenum (61%), tungsten (57%), and selenium (46%). ₂ The digest summarizes these as 47-61%, which conflicts with its own listed selenium value of 46%. It also judges the experiment-count bias to be real but less severe than initially thought. ₂

The overlap analysis cannot by itself distinguish chloride, sodium, osmotic stress, or other shared physiological effects, because NaCl contains both Na⁺ and Cl⁻. ₁ No hypergeometric functional-enrichment test was performed on the shared-stress versus metal-specific classification, and the SEED annotation comparison was descriptive. ₁ Unmeasured essential genes therefore cannot be assigned confidently to either shared-stress or stressor-specific fitness requirements. This connects the decomposition problem to essentiality-assay-discordance and gene-essentiality. ₁

Tensions

The two analyses report different per-metal conservation estimates. After shared-stress correction, the counter-ion analysis reports manganese at +0.182, zinc at +0.115, and iron at +0.182. Its own uncorrected values were manganese +0.182, zinc +0.145, and iron -0.040. ₁ The primary atlas’s original, uncorrected estimates are manganese +0.198, zinc +0.151, and iron +0.116. ₅ The discrepancy is therefore present even before any shared-stress treatment. The values should not be reconciled by averaging, because the analyses differ in record sets, organism coverage, and procedure. Matched-data reanalysis is required to determine whether the discrepancy is methodological or biological. _1,5

The metal cross-resistance study interprets its metal-pair correlations as a universal positive directional layer. Within that report, the statement that all tested metal pairs are positive sits alongside a count of 311/317 (98.1%) positive correlations. These are conflicting source statements, and universal positivity should not be treated as established. ₄ The counter-ion analysis, by contrast, shows that metal–NaCl correlations vary substantially across metals, including iron r=0.086 and zinc r=0.715. ₁ This is not a direct contradiction, because the analyses compare different condition pairs. It does refine the interpretation: predominantly positive metal–metal correlations do not establish a uniform response to NaCl or identify which shared stressor generates the signal. _4,1

The projects also disagree on how much of the metal-important set reflects general stress. The atlas’s limitations section describes most genes captured by its broad metal-important rule as general stress genes. By contrast, the counter-ion analysis classifies 60.2% of metal-important records as metal-specific relative to NaCl. The discoveries digest concludes from the metal-specificity classification that the majority (55%) are genuine metal tolerance determinants. _5,1,2 These fractions rest on different importance definitions, comparison conditions, and record sets. The disagreement should be kept open rather than settled by preferring one project. _5,1,2

Open Directions

• Compare metal–choline chloride (ChCl) overlap with metal–NaCl overlap to separate chloride from osmotic confounding. Several Fitness Browser organisms already have ChCl experiments, though at different concentrations. The report assumes ChCl provides Cl⁻ without Na⁺/osmotic effects, but that assumption is untested here, and how fully ChCl isolates osmotic effects remains uncertain. Add matched KCl controls profiled with RB-TnSeq to test whether shared fitness signals track chloride, sodium, osmolarity, or the combined salt condition. This comparison is proposed, not completed. ₁
• Run the report’s highest-impact proposed follow-up. This would be RB-TnSeq with matched metal chloride and metal sulfate concentrations in a single organism under identical growth conditions, removing the aerobic/anaerobic confound that limits the psRCH2 comparison. For copper, the most-tested metal, the proposed pair is CuCl₂ 0.5 mM versus CuSO₄ 0.5 mM in MR-1 or DvH. Zinc has already been tested as sulfate, so adding ZnCl₂ 0.2 mM against ZnSO₄ 0.2 mM in MR-1 would directly test the null. Cobalt is described as having the highest Cl⁻ load in the current data, so comparing CoCl₂ 0.1 mM with CoSO₄ 0.1 mM in DvH would test the mechanism. These are proposed conditions, not reported results, and each comparison should hold oxygen regime, media, and culture conditions identical. ₁
• Generate DvH NaCl dose-response data at 0.1, 1, 10, 100, and 500 mM to dose-match NaCl stress with the effective chloride delivered by metal salts. This would test whether the metal–NaCl correlation hierarchy changes with dose. These experiments are proposed, not reported as completed. ₁
• Use COG (Clusters of Orthologous Groups), KEGG, and PFAM annotations to formally test two predictions. The first is that shared-stress genes are enriched for envelope/transport functions; the second is that metal-specific genes are enriched for metal-binding/efflux. The test would also cover the proposed cell-envelope, DNA-repair, and ion-homeostasis categories. This enrichment remains a proposed test, not an established result. ₁
• Apply independent component analysis (ICA), a method that decomposes correlated measurements into latent components. Also run the metal–NaCl overlap analysis at the module level, using modules from the fitness_modules project. Whether metal-responsive modules overlap NaCl-responsive modules remains an open question. ₁ The metal-specificity analysis identified 0 metal-specific modules using per-module z-normalization. Its raw module-condition scale, however, differed from the precomputed z-scored profiles that identified 600 metal-responsive module records. Rerunning the comparison on the latter profiles would test whether the negative result is scale-dependent. ₆
• Reanalyze the 10,821 metal-important records with stricter and more permissive NaCl-importance thresholds. Normalize the n_sick rule for NaCl experiment count, for example SynE’s 12 experiments versus 1–6 elsewhere. Combine this with condition-specific models using matched salt controls to quantify how classification rules affect the 39.8% shared-stress fraction. _1,2
• Resolve the metal-specificity locusId mismatches and repeat its classification across the excluded organisms. This would test whether attrition, and DvH’s 608 non-metal experiments, change the shared-versus-specific balance and the essential-metal specificities. _6,2
• Recompute the counter-ion and atlas conservation statistics on identical organism, metal, and gene-record sets. This would resolve the reported manganese, zinc, and iron discrepancies, including those in the uncorrected baselines, and the cadmium/iron sign-change wording. _1,5
• Combine matched salt-control fitness profiles with functional annotation and expression measurements. This would test whether the 149 conserved, incompletely annotated metal-biology candidates are shared-stress or stressor-specific. Start with the digest’s metal-specific candidates UCP030820, YebC, and DUF1043-YhcB. _5,2
• Apply PGLS or independent contrasts to the cross-resistance matrices, normalize metal concentrations relative to MIC, and add non-metal stress controls. This would determine whether positive cross-resistance reflects universal stress biology or conserved metal-pair chemistry. ₄
• Map cross-resistance gene signatures with KEGG/PFAM across 27K species and compare them with matched salt-control profiles. This would test whether the three gene tiers predict metal-specific rather than merely shared-stress fitness. ₄
• Separate the top 18 housekeeping amino-acid-biosynthesis genes from the ~370 condition-restricted genes in the FW300 metabolite data before Web of Microbes and Fitness Browser integration. This would test whether substrate-specific concordance improves once shared housekeeping fitness is removed. ₇
• Validate the metal-specificity classification against the Fitness Browser specificphenotype annotations. Also rerun it with the counter-ion threshold (fit < -1 without |t| > 4). Together these would test whether the 2.7x osmotic-overlap discrepancy is purely methodological. ₆
• Repeat the atlas conservation analysis using only genes important for metals but not for other stresses, as the atlas proposes. This would separate the core stress-response signal from metal-resistance gene evolution. ₅
• Fit the soil project’s pending partial-correlation model (COG ~ Cr | Cu + Zn + Pb). Then compare the COGs that survive conditioning with the orthologs of fitness-defined metal-specific genes. This would test whether field metal associations separate into shared and metal-specific components as fitness data do. ₉