GC mid2

Deep research

Generated by just research-trait; source: research/traits/genomics/gc_mid2-deep-research-falcon.md

# Curation-focused research report: GC mid2

## Executive assessment

**Target trait:** GC mid2  
**Trait CURIE:** `METPO:1000431`  
**Category / kind:** GENOMICS / CLASS  
**Operational definition supplied:** genome-wide GC composition of approximately **57.0–66.3%**, corresponding to `GC_57.0_66.3`.

GC mid2 is best modeled as an **assembly-level compositional phenotype**, not as a physiological activity, environmental preference, or single-gene phenotype. Genomic GC content is the fraction of called DNA bases that are guanine or cytosine, normally calculated as `(G+C)/(A+T+G+C) × 100`. It is a slowly evolving outcome of mutation spectra, DNA replication and repair, recombination-associated fixation bias, selection, drift, and genome history. Recent large-scale work emphasizes strong phylogenetic inertia: among 11,083 representative bacterial genomes, genomic GC ranged from about 16% to 77%, had a bimodal distribution, and more than 60% of variance was explained at the phylum level; Blomberg’s K was 1.47 and Pagel’s λ was 0.998. Most genomes clustered below 45% or above 60%, placing much of GC mid2 within the lower portion of the high-GC mode rather than at a universal biological threshold. (teng2023genomiclegaciesof pages 2-5)

The most defensible core graph is therefore:

> **DNA damage/replication errors → nucleotide-specific mutation spectrum → G/C-versus-A/T mutation pressure**, opposed by **homologous recombination → GC-biased gene conversion → preferential fixation of G/C alleles**, with DNA-replication/repair-system state modifying these rates. Long-term balance among these processes can produce a genome in the 57.0–66.3% bin. (teng2023genomiclegaciesof pages 2-5, teng2023genomiclegaciesof pages 8-10, lassalle2015gccontentevolutionin pages 4-6)

No study identified a mechanism uniquely producing the **57.0–66.3% interval**. The bin is an ontology discretization of a continuous variable. Mechanistic edges should consequently terminate in “increased/decreased genomic GC content” before a final threshold-classification edge to `METPO:1000431`.

## 1. Trait scope and boundary cases

### Included phenotype

The trait represents the **whole-genome or assembly-wide nucleotide fraction** lying between approximately 57.0% and 66.3% GC. Coding-sequence GC can be used as a proxy only when whole-genome GC is unavailable: in a 2023 compilation of 49,783 prokaryotes, coding-sequence and genomic GC were nearly identical at population scale (`Spearman r=0.99`, adjusted `p<9×10⁻²⁰⁰`). That study found a total prokaryotic range of 16–77%, with 90% between 33% and 71%. (aliperti2023rkselectionof pages 3-6, aliperti2023rkselectionof pages 1-3)

### Exclusions and neighboring properties

1. **Not GC3:** GC at synonymous third-codon positions is a related but distinct measurement and can respond more strongly to recombination, codon selection, and mutation bias. It must not be substituted for genome-wide GC. In bacterial comparative data, recombinant genes often showed a larger GC difference at GC3 than across all codon positions. (lassalle2015gccontentevolutionin pages 4-6)
2. **Not local GC enrichment:** Individual genomic islands, horizontally acquired genes, recombination tracts, restriction sites, or high-GC genes do not establish the assembly-level trait.
3. **Not merely “high GC”:** The interval includes moderately high through high-GC genomes but excludes genomes above approximately 66.3%. For example, the reported 66.61% GC of *Deinococcus radiodurans* lies just above the supplied upper boundary, whereas *Pseudomonas fluorescens* at 60.50% is within it. (long2018specificityofthe pages 1-2)
4. **Boundary uncertainty:** The METPO source should be checked to determine whether 57.0 and 66.3 are inclusive and how rounding is handled. A genome reported as 56.96%, 57.04%, 66.29%, or 66.34% can change class under one-decimal rounding.
5. **Assembly bias:** Contaminated, incomplete, plasmid-enriched, or metagenome-assembled genomes can have biased GC estimates. Curators should record assembly scope and calculate GC from comparable nucleotide classes.
6. **Organellar and viral sequences:** Unless METPO explicitly says otherwise, plasmids, phages, mitochondria, chloroplasts, and other replicons should not be pooled indiscriminately with the principal microbial chromosome.

### Recommended terminal graph representation

- `genomic GC content` — **has quantitative value** → `57.0–66.3 percent`
- `microbial genome` — **has trait** → `METPO:1000431`
- `increased genomic GC content` — **may result in threshold membership** → `METPO:1000431`

The last edge should be interpreted as classification, not biological causation.

## 2. Current mechanistic understanding

### Mutation pressure

Comparative bacterial work reports a pervasive excess of **G/C→A/T mutations**, even in high-GC genomes. Mutation pressure alone would therefore generally pull genomes toward lower equilibrium GC, requiring an opposing fixation process to explain observed compositions substantially above the mutation-only equilibrium. (lassalle2015gccontentevolutionin pages 1-4, lassalle2015gccontentevolutionin pages 4-6)

Relevant chemical routes include cytosine deamination, which converts a GC-base-pair state toward an AT state if unrepaired, and oxidation of guanine to 8-oxo-guanine, which can mispair with adenine during replication. The 2023 synthesis explicitly states that cytosine deamination “reduces the GC content” and that oxidized guanine produces A–8OG mispairing. (teng2023genomiclegaciesof pages 8-10)

Mismatch repair is a major modifier of mutation rate and spectrum rather than a universal “GC-raising enzyme.” In mutation-accumulation experiments, deleting `mutS` increased the *P. fluorescens* base-substitution rate approximately 309-fold, while deleting `mutL` increased it approximately 278-fold. Repair specificity also depends on neighboring base composition. These experiments establish that MMR strongly changes mutational input, but they do not establish that MMR always drives genomes into or out of GC mid2. (long2018specificityofthe pages 1-2)

### Recombination and GC-biased gene conversion

In bacteria, homologous recombination can produce gene conversion, a unidirectional transfer from donor to homologous recipient. GC-biased gene conversion (gBGC) denotes preferential transmission or fixation of G/C over A/T alleles during resolution of mismatches in recombination tracts. Unlike adaptive selection, gBGC can increase GC even if the G/C allele has no fitness advantage. (lassalle2015gccontentevolutionin pages 1-4, lassalle2015gccontentevolutionin pages 4-6)

Across 20 bacterial groups and one archaeal group, seven groups were too clonal for informative analysis because fewer than 2% of tested core genes showed recombination. In 11 of the remaining 14 groups, recombinant genes had significantly higher total GC and/or GC3; the effect was consistently stronger at GC3. Exceptions included *Helicobacter pylori* and the *Bacillus anthracis/cereus* group. This is substantial comparative support for bacterial gBGC, but it remains indirect and is not evidence that gBGC alone fixes a genome specifically in the GC-mid2 interval. (lassalle2015gccontentevolutionin pages 4-6)

### DNA replication and repair system architecture

The most important recent mechanistic-development paper is Teng et al. (February 2023). In 11,083 representative bacterial genomes, a phylogenetically controlled model based on 217 DNA-replication-and-repair KEGG orthologs explained up to **88% of total GC variance**, with multiple correlation coefficient 0.94. DnaE2, an error-prone translesion-synthesis polymerase, had the strongest positive association, whereas MutS2 had the strongest negative association. Positively associated proteins were enriched in base-excision repair, NHEJ, translesion synthesis, and nucleotide-excision repair; several MMR, homologous-recombination, and replication proteins were negatively associated. (teng2023genomiclegaciesof pages 5-8, teng2023genomiclegaciesof pages 2-5)

Showing the first 60 of 250 lines of findings; the linked file also carries the run's front matter and the prompt it was given — read the full report.