The Distribution of Fitness Effects of Mutations in

Humans and Flies

Adam Eyre-Walker

(University of Sussex)

Types of Mutation

• Deleterious

• Neutral

• Advantageous

0 +ve-ve

Deleterious Mutations

1/100

<10%

1/10,000

<30%

Mutationaccumulation

and Mutagenesis expts

dn/ds

in primates

Distribution of Effects

neutraldeleterious low high

Theory

Neutral sites (e.g. introns / synonymous)

Selected sites (e.g. non-synonymous)-assume all mutations neutral or deleterious

Ps=Ls i i1

i =1

n - 1

!

Pn=Ln i D(S) x(1 - x)(1 - e- S)(1 - e- S(1 - x) )## (1 - xn - (1 - x)n)2S2x

Simplication

f +Ne

b1

Theory

Neutral sites

Selected sites

Ps=Ls i i1

i =1

n - 1

!

Pn . Ln i i1

i =1

n - 1

! ai b1

Parametersn - knownLn - each geneLs- each gene - each gene - shared - shared

Estimationassume free recomb, , Bayesian estimationusing MCMC

Dataset - humans

• Environmental genome project

• 275 human genes

• 90 individuals resequenced

• 549 non-synonymous polymorphisms

• 15746 intron polymorphisms

Pn/Pi versus i

Human

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

-0.002 0.002 0.006 0.01 0.014

Pn/Pi

i

Results - human

Nes 01 110 10100 1001000 100010000

% 23 22 37 19 0.1

Shape = 0.28Nes = 240

Results - human

01 110 10100 1001000 100010000

0.38 0 0 0 0.62

0.23 0.22 0.37 0.19 0.001

0.17 0.33 0.47 0.03 0.000

Shape = 0.28 (0.03, 0.48)

Nes = 240 (90, )

Low Frequency Polymorphisms

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

low medium high

frequency

proportion

synnon-syn

Dataset - D.melanogaster

44 genes 5-55 alleles sequenced 141 non-synonymous polymorphisms 346 synonymous polymorphisms

Pn/Ps versus s

D.melanogaster

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

-0.01 0 0.01 0.02 0.03 0.04 0.05

Shape = 0.46 (0.15, 0.65)

Adaptive Mutations

Human1 CCC GCA GAG TTA CTA ATC GAAHuman2 CCG GCA GAG TTA CTA ATC GAAHuman3 CCC GCA AAG TTA CTA ATC GAAHuman4 CCC GCA AAG TTA CTA ATC GAA

Chimp CCC GCC GAG TTA GTA ATT GAA

Poly Sub

Syn 1 2

Non 1 1

Model

Poly Sub

Syn Ps≈4Neu Ds≈2ut

Non Pn≈4Ne u f Dn≈2 ut f + a

Assume - synonymous mutations are neutral - amino acid mutations are deleterious, neutral or advantageous

a=Dn - Ds Ps

Pn

a =1- DnPs

DsPn

Estimation

Parameters

n, Ln, Ls - known without error - each gene - each gene - shared, beta distributed or one per gene

Estimation by ML

Drosophila

Poly Sub

Syn 707 2489

Non 153 1054

0.22 0.42

35 genes with multiple alleles in D.simulans and one allele in D.yakuba

Result

= 0.26 (0.08, 0.41)

Proportion Constant

Gene Amino Acid Div

Hsc70 0.0023

Adh 0.036

Est-6 0.20

Model n Log(L)

One 106 -327.5

Beta distributed 107 -327.5

One per gene 140 -302.9

D.simulans & D.yakuba

600,000 aa differences

26 % adaptive

160,000 adaptive

1 every 75 years

Human-Chimp

• Environmental Genome Project• 232 human genes• 90 individuals resequenced• Non-synonymous versus intron• Human sequence aligned against

chimpanzee genome

Human Nuclear Genes

Poly Sub

Intron 17631 33223

Non 681 765

0.039 0.023

Low Frequency Polymorphisms

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

low medium high

frequency

proportion

synnon-syn

Dealing With Deleterious Mutations

• Use estimate of distribution of fitness effects from SNP data

• Assume adaptive and slightly deleterious mutations governed by one distribution

• Ignore low frequency variants

Excluding SNPs

Cutoff ML 95% CI

0% -0.62

5% 0.09 (-0.11, 0.26)

10% 0.26 (0.08, 0.41)

20% 0.31 (0.11, 0.52)

Humans & Chimpanzees

1%290,000 amino acid differences

25% adaptive72,500 adaptive differences

1 every 165 years

Conclusions

• Distribution of fitness effects of slightly/moderately deleterious mutations is highly leptokurtic in humans and drosophila

• ~25% of amino acid substitutions are driven by positive selection in humans and drosophila

• Proportion does not vary between genes

Thanks

Nick Smith

Nicolas Bierne

Gwenael Piganeau

Meg Woolfit