Detecting Adaptive Evolution

Alan R. Rogers

November 14, 2018

Conventional wisdom

Something must have happened to weaken the selective pressuredrastically. We cannot escape the conclusion that man’s evolutiontowards manness suddenly came to a halt.

—Ernst Mayr 1963

Natural selection has almost become irrelevant in human evolution.There’s been no biological change in humans in 40,000 or 50,000years. Everything we call culture and civilization we’ve built withthe same body and brain.

—Stephen Jay Gould 2000

Certainly, human nature is fixed. It’s universal and unchanging—common to every baby that’s born, down through the history ofour species.

—Helena Cronin 2000

Is this really true? How could we know?

Signatures of selection

I High proportion of functional changes

I Reduction of gene diversity

I Population differences

I Excess of singletons

I Common allele on long LD block

Time scale for signatures of selection

PRM1 gene

I compacts sperm DNA

I 13/14 human-chimp diffs are non-synonymous (6 shown here)

Proportion of functional changes at FOXP2 locus

I Mutations at FOXP2 cause problems with language.

I Few amino acid changes within mammals ⇒ strong selectiveconstraint.

I Two mutations on human lineage, neither synonymous ⇒selection

(Enard et al 2002)

Evolution of FOXP2

Vertical bars separate nucleotide substitutions; gray indicates amino-acidchanges. (Enard et al 2002)

Melanocortin 1 receptor (MC1R) locus

I Affects color of skin and hair.

I human-chimp Ka/Ks = 0.63: large for a functional protein ⇒weak selective constraint

I Yet Ka/Ks = 0 among Africans.

I On the other hand, Ka/Ks � 0 among Europeans

Hypothesis

I Before loss of body hair: weak selective constraint on MC1R

I After loss: constraint strong w/i Africa; weak w/i Europe

I Selective sweep at loss of body hair

Time required to generate African neutral π̂

π

0 0.5 1 1.5Millions of Years

π̂ = .67

←3 half-lives = 1.2 myr

N =∞↘

....................................................................................................................................................................................................................................................................................................................................................

↖N = 14000

......................................................................................................................................

....................................................

.........................................................................

...............................................................................................................................................................................

↖N = 30000

............................................................................................................................................................................................................................................................................

..............................................

..................................................

...................................................

Trajectories of π following a selective sweep. Dotted lines show thehalf-lives when N = 14, 000; the dashed line shows π̂, the observed valueof π.

Duffy map How a selective sweep generates LD

Huge block of LD around lactase allele in Europe Huge block of LD around lactase allele in Europe

(Nathan Harris)

Signature of an ongoingselective sweep at DRD4I Sweeping allele is

I commonI has low diversity

over large region

I High LD over largeregion

Linkage disequilibrium at G6PDLeft: LD plot; right: haplotype plot

Study of Voight et al (2006)

I 800,000 SNPs in 309 people

I 431 sweeping loci

I Most sweeps started w/i past 10,000 years

LD on human chromosome 2 (Voight et al 2006)

Voight et al (2006):431 sweeping loci.

ASN: AsiaYRI: AfricaCEU: Europe.

Most are sweeping w/ionly one continent.

Gene diversity vs. distance from exons, scaled byhuman-rhesus divergence

Hernandez et al (2011)

Gene diversity vs. distance from exons, scaled byhuman-rhesus divergence

Hernandez et al (2011)

Diversity is low near exons. Why?

Old selective sweeps?

Background selection (againstdeleterious mutations)?

If it was sweeps, effect should be largest near exons withhuman-specific amino-acid substitutions.

But the patterns around synonymous and non-synonymoussubstitutions are the same.

Hernandez et al (2011)

But the patterns around synonymous and non-synonymoussubstitutions are the same.

Hernandez et al (2011)

If selective sweeps caused the dipin diversity, we would expect thedip surrounding non-synonymoussubstitutions to be wider.

This pattern is seen in Drosophilasimulans, but not in humans.

Suggests that classical selectivesweeps play only a minor role inadaptive evolution amonghumans.

Alternatives: selection from standing variation, and selection onquantitative variation. We need tools to study these effects.

The Singleton Density Score (SDS, Field et al 2016)

1. The environment changes, some class of haplotypes becomesadvantageous, and begins to grow.

2. Growth within this class ⇒ recent coalescent events ⇒ shortterminal branches ⇒ few singleton mutations.

3. Class of disfavored haplotypes: long terminal branches ⇒more singleton mutations.

SDS is a normalized estimate of the difference in tip lengthbetween the haplotypes linked to two alleles at a nucleotide site.

Large samples provide sensitivity to recent selection. A sample of3000 provides sensitivity to selection during ∼75 generations,∼2000 y.

Selection on height alleles in UK

X axis: strength ofGWAS evidence thatlocus affects stature

Y axis: strength ofselection for “tall”

allele.

Many of these GWASassociations are not

statistically significant.Yet in aggregate, they

demonstrategenome-wide selection

increasing stature.

Summary

I Signature of a recent classic selective sweep: common derivedallele surrounded by extensive LD.

I Many such sweeps have been discovered.

I Tend to be population-specific.

I In humans, classic sweeps account only for a minority ofadaptive evolution.

I A new method (SDS) makes it possible to study recentselection on standing variation and on polygenic characters.