Statistical Genomics Foundry
Paper
domain/molecular-evolutiontopic/gc-biased-gene-conversion

Galtier & Duret 2007 — Adaptation or biased gene conversion?

1. Citation

Galtier N, Duret L. 2007. “Adaptation or biased gene conversion? Extending the null hypothesis of molecular evolution.” Trends in Genetics 23(6):273–277. Epub 2007 Apr 5. DOI 10.1016/j.tig.2007.03.011. PMID 17418442. Author affiliation: CNRS UMR 5554 — Institut des Sciences de l’Evolution, Université Montpellier 2. No open-access full text located. Access date: 2026-07-05.

2. Access note

Abstract-only. The full text is behind the Elsevier / Trends in Genetics paywall (sciencedirect.com returned HTTP 403; cell.com and Elsevier landing pages are gated). The only portion actually read is the NLM/PubMed abstract (verbatim via NCBI eutils efetch, PMID 17418442) plus the PubMed record’s metadata. Everything below the abstract line — the body text, figures, the mouse Fxy worked example, and the enumerated discrimination criteria — was NOT accessible and is not summarized here. Where the abstract is silent, this note says so rather than filling in.

3. Thesis (1 sentence)

Accelerated molecular evolutionary rates, ordinarily read as a signature of natural selection, can instead be produced by GC-biased gene conversion, so gBGC should be treated as an extended null hypothesis to be excluded before inferring adaptation.

4. Problem & context

Analyzing evolutionary rates is a common way to detect natural selection at the molecular level: the expectation is that sequences contributing to a species’ adaptation evolve faster than nonfunctional sequences, because favorable mutations have a higher fixation probability than neutral ones. The paper’s problem is that an elevated rate is not uniquely diagnostic of selection — a non-selective process (GC-biased gene conversion) can generate the same acceleration signal.

5. Method / approach

[From the abstract only — the body-level method was not accessible.] The paper is a review/opinion-style Trends in Genetics article. It (a) argues that accelerated evolution can arise from GC-biased gene conversion rather than selection, applying not only to neutral sequences but also to constrained (functional) sequences; (b) uses the mouse Fxy gene as an illustrative case of a constrained sequence affected this way; and (c) states that “several criteria” can discriminate the natural-selection model from the biased-gene-conversion model, then applies them to the recently reported human accelerated regions. The abstract does not enumerate or name those criteria, and the enumerating body text was paywalled, so the specific discriminators are not recoverable from what was read.

6. Key claims / findings

  • An accelerated substitution rate can be caused by factors other than natural selection, in particular GC-biased gene conversion. [abstract]
  • This confound applies both to neutral sequences and to constrained (selectively functional) sequences — illustrated with the mouse Fxy gene. [abstract]
  • Several criteria can discriminate between the natural-selection model and the biased-gene-conversion model. [abstract] — the criteria themselves are not stated in the readable portion.
  • Applying those criteria, the recently reported human accelerated regions (HARs) are most likely the result of biased gene conversion rather than adaptation. [abstract]
  • Framing consequence: such regions, rather than contributing to human adaptation, might represent “the Achilles’ heel” of the genome (i.e., a vulnerability, not an advantage). [abstract]

7. Load-bearing statements — OWN-WORDS (restrictive license: Elsevier, all rights reserved)

Mode: the source is all-rights-reserved (Trends in Genetics / Elsevier), so load-bearing statements are paraphrased, not quoted. No functional strings requiring verbatim reproduction (e.g. W→S / S→W substitution-class labels, numeric thresholds) appear in the readable abstract.

  • Rate analysis is a popular way to characterize selection’s effect at the molecular level; adaptive sequences are expected to evolve faster than nonfunctional ones because favorable mutations fix more often than neutral ones. (own-words paraphrase of abstract)
  • An accelerated rate “might be due to factors other than natural selection, in particular GC-biased gene conversion” — true for neutral and also for constrained sequences. (own-words paraphrase)
  • The reported human accelerated regions are, on these criteria, most likely a product of biased gene conversion, not adaptation. (own-words paraphrase)

8. Stated scope, assumptions, limitations

[From abstract only.] Scope is the interpretation of accelerated evolutionary rates and specifically the human accelerated regions; the Fxy gene is the constrained-sequence example. The abstract does not state quantitative thresholds, sample sizes, or the boundary conditions under which each discrimination criterion holds — those (if present) are in the paywalled body.

9. Failure modes / invalidity patterns

[From abstract only.] The invalidity pattern the paper names is at the level of inference: attributing an accelerated evolutionary rate to positive/natural selection when the acceleration was actually produced by GC-biased gene conversion. gBGC is presented as an alternative (non-adaptive) explanation that must be excluded — an extended null hypothesis. The abstract asserts that discrimination is possible via “several criteria” but does not spell out the diagnostic symptom, direction, or any threshold value in the readable text. The specific detectors (e.g. any substitution-asymmetry statistic or recombination-rate correlation) are not recoverable from the abstract.

10. What the source does NOT address (in the readable portion)

  • The mechanistic detail of gBGC (preferential fixation of W→S vs S→W alleles, high-recombination regions) is not described in the abstract.
  • The identity of the “several criteria” is not given in the abstract.
  • Any numeric threshold or ratio (e.g. a W→S : S→W cutoff) does not appear in the abstract.
  • Correlation with local recombination rate is not mentioned in the abstract. (These may well be in the paywalled body; this note cannot confirm either way.)

11. Open questions / ambiguities

  • What are the “several criteria”? Not recoverable here — requires the full text.
  • Does the full paper state a quantitative discriminator (e.g. a substitution-class ratio or a recombination correlation coefficient/threshold)? Unknown from the abstract.

12. Guidance answers

  • gBGC mechanism (W→S / A/T→G/C preferential fixation in high-recombination regions, independent of selection, and why codon models misread it): NOT stated in the readable portion. The abstract names “GC-biased gene conversion” as the confounding process and says it can accelerate both neutral and constrained sequences, but it does not describe the W→S allele-fixation mechanism, the recombination dependence, or the codon-model misattribution pathway. That detail is in the paywalled body. (The substitution-class labels W→S / S→W — functional strings that would be kept verbatim — do not appear in the abstract.)
  • gBGC framed as an alternative null hypothesis to test against selection: YES, at the framing level. The title and abstract present gBGC as “extending the null hypothesis of molecular evolution” — i.e. a non-adaptive alternative that must be excluded before concluding selection. (Exact wording is paraphrased per the restrictive-license mode; the title phrase “Extending the null hypothesis of molecular evolution” is the paper’s own title.)
  • Named diagnostic + numeric threshold (incl. the external “W→S ratio > 1.5” attributed to “Galtier 2013”): The abstract states only that “several criteria” discriminate the two models; it names none and gives no numbers. This 2007 paper’s readable portion states no “W→S ratio > 1.5” or any other numeric threshold. Per guidance: that cutoff is therefore NOT sourced from here — treat it as convention or as belonging to a different (later) reference, not to Galtier & Duret 2007. Whether the paywalled body of this paper lists specific criteria is unconfirmed.