BGPT: Paper Review: Population Genomics of Human Adaptation

Explore by Goal

Quick Explanation Copied

Paper review (rigorous + skeptical) — Population Genomics of Human Adaptation

A high-signal review that (1) organizes the main selection-scan statistics (dN/dS, MK, EHH/iHS, Tajima’s D/Fay & Wu’s H/MFDM, FST/outliers, environmental association, QST–FST, CMS integration), (2) repeatedly emphasizes demography-driven false positives (bottlenecks, allelic surfing, biased gene conversion), and (3) grounds several examples (immune genes, reproduction, lactase persistence, EDARV370A, FOXP2, high-altitude adaptation) in the need for functional validation beyond statistics.

Primary review DOI:

Long Explanation

BGPT Visual Paper Review

Paper: Population Genomics of Human Adaptation — Lachance & Tishkoff (2013, Annual Review of Ecology, Evolution, and Systematics).

What you can extract

Method map: comparative genomics (dN/dS, MK), haplotypes (EHH, iHS), allele frequency spectra (Tajima’s D, Fay & Wu’s H, MFDM, PRF), multi-population signals (FST/outliers, XP-CLR, LSBL/PBS), environmental associations (Bayes factors vs neutral covariance), and integrative CMS.
Major confounders: demography mimicking selection; limited power for soft sweeps; ascertainment bias; multiple testing; and non-adaptive explanations such as allelic surfing and GC-biased gene conversion.
Examples: immune and reproduction signals genome-wide; lactase persistence (convergent evolution); EDARV370A (eccrine/hair/tooth pleiotropy); FOXP2 signals; and high-altitude adaptations (Tibetan/Andean/Ethiopian patterns), each tied to the need for physiological follow-up.

1) Visual overview: “signatures of selection” the review highlights

The review frames signatures of adaptation as including sequence conservation changes, reduced polymorphism, shifted allele-frequency distributions, extended haplotype homozygosity, and population-differentiation outliers.

2) Quantitative anchors found in the text (used to build graphs)

The review reports specific quantitative takeaways from cited selection-scan analyses. Below, those numeric statements are extracted verbatim from the provided text (no invented values).

“9%” of informative loci show positive selection signals in a framework contrasting polymorphism and divergence (Bustamante et al. 2005 cited in the review).
“8%” upper bound for proportion of positively selected mutations in a set of 50 genes, from a PRF-based analysis (Nielsen et al. 2005 cited in the review).
Immune signals are described as prominent genome-wide targets of positive selection.

Skeptical note: these percentages are not directly comparable because the second estimate is explicitly framed as an upper bound due to focusing on loci with already-strong signals.

3) Method “decision tree” (built from the review’s organization)

Below is a conceptual graph that organizes which data type supports which statistical signal—as explicitly discussed in the review.

The review explicitly covers comparative genomics, haplotype approaches, allele-frequency distributions, multi-population data, environmental integration, QST–FST comparisons, and integrative CMS.

4) Critical evaluation: strongest contributions vs key fragilities

Strengths

Methodological breadth with explicit coverage of major statistic families and how they relate to sweep timing and data structure (comparative vs haplotype vs allele-frequency vs multi-population vs environmental correlation vs phenotype-based comparisons).
Explicit confounder discussion—the review repeatedly addresses that demography can mimic selection and that not all signals correspond to adaptation. It highlights allelic surfing and GC-biased gene conversion as non-selective processes that can resemble selection patterns.
Integration + validation emphasis: CMS is presented as combining multiple signals, and the review concludes that selection scans require functional follow-up to demonstrate physiological impact.

Fragilities / blind spots (within the review’s own framing)

Demography vs selection separation remains “hard”: the review notes that population size changes and bottlenecks distort allele frequency spectra and that haplotype statistics can be affected by recombination hot spots, recombination-rate differences, and limited power for soft sweeps.
Power and error-control: the review explicitly cautions about multiple testing over millions of SNPs and highlights the importance of multiple-comparison correction and ascertainment bias in SNP arrays/ascertained SNP datasets.
Functional annotation isn’t a substitute for causality: even if loci replicate, the review notes that adaptive variants must change fitness, and it warns against “adaptive storytelling” (Panglossian Paradigm) and against over-interpreting functional annotation without validation.

5) Worked “example loci” table from the review (no new claims)

Adaptive signal example	What the review says was found	Key caveat mentioned
Immune-related genes	Immune-related genes are “major targets of positive selection” genome-wide.	Selection-scan outputs require functional follow-up to demonstrate physiological impact.
Lactase persistence (LP)	LP variants show convergent evolution; European and African-associated mutations occur on distinct haplotype backgrounds; sweeps differ in timing (review cites differing EHH lengths and timing ranges).	Environmental/cultural context affects interpretation; and statistical sweeps can be confounded by demography or non-adaptive processes.
EDARV370A / eccrine glands	EDARV370A is described as a strong selection signal in Asian populations; pleiotropic effects make it unclear which trait selection targeted.	Pleiotropy means genotype–phenotype mapping is not one-to-one; causal trait under selection may remain ambiguous.
FOXP2 and language	FOXP2 is presented as undergoing positive selection signals; review discusses derived substitutions and timing hypotheses tied to allele-frequency/haplotype patterns.	Complex selection timing signals require careful dating and demography-aware interpretation.
High-altitude adaptation	Review summarizes genetic signatures in Tibetan (e.g., EGLN1, EPAS1) and contrasts with other populations (Andean, Ethiopian), noting genetic architecture variability.	Same phenotype can arise through different genetic architectures; correlational signals may not reflect causal physiology without follow-up.

Every row above is grounded in the review’s own summaries and cautions.

6) “What would disprove this?” (falsification-oriented critique)

Because this is a review, “disproof” targets the inference pipeline used by studies it synthesizes: selection-scan candidates may not reflect fitness changes, may be artifacts, or may not replicate across populations or methods.

Adaptive storytelling risk: the review highlights Gould & Lewontin’s critique as a warning that apparent adaptation may be inferred without adequate causal demonstration.
Spurious selection signatures: it explicitly mentions allelic surfing and GC-biased gene conversion as processes that can mimic positive selection-like patterns.
Statistical fragility: it also notes limitations in multiple testing and SNP ascertainment bias which can produce false positives.

7) Author conflict-of-interest statement (from the provided text)

The review includes a disclosure statement: the authors report not being aware of affiliations/funding/holdings that might affect objectivity.

Next: bespoke BGPT author reviews

Click to open focused critiques/summaries by each full author name.

Feedback:

Updated: March 21, 2026