Document 11743575

advertisement
9/16/14 Chapter 2 – Genetic Drift & Mutation
The Wright-Fisher Model
Mutation
The Wright-Fisher model
sample 2N gametes
∞
2N haploid
individuals
gametes
2N
individuals
p0
p0
p1
v effectively a haploid model, but with population
size = 2N to mirror a diploid population
v random sampling of genotypes and random
sampling of gametes essentially the same
http://psych.colorado.edu/~carey/hgss/hgssapplets/evolution/geneticdrift/GeneticDrift.html
http://darwin.eeb.uconn.edu/simulations/drift.html
1 9/16/14 Expectation
v What is the expected payoff from buying
a $1 lottery ticket?
Expectation
v What is the expected payoff from buying
a $1 lottery ticket? ~$0.18 + JACKPOT
Match
5
5
4
4
3
2
3
1
0
+ Megaball
+
1
+
1
+
+
1
1
+
+
1
1
Probability
1 in 175,711,536
1 in 3,904,701
1 in 689,065
1 in 15,313
1 in 13,781
1 in 844
1 in 306
1 in 141
1 in 75
Top Prize
JACKPOT
$250,000**
$10,000**
$150**
$150**
$10
$7
$3
$2
2 9/16/14 Expectation
v What is the expected payoff from buying
a $1 lottery ticket? ~$0.77
Massachusetts
Net Ticket
Sales*
Prizes*
Prizes as %
of Ticket
Sales
4,200
3,235
77.00%
Admin
Costs*
83
Admin as
Net
% of Ticket Proceeds
Sales
to State*
2.00%
Proceeds
as % of
Ticket
Sales
883
21.00%
* values in millions
Expected Allele Frequency
v the expected allele frequency in the next
generation is equal to the current allele
frequency, but that’s only what we expect on
average
E !" f A (t +1)#$ = 2Nf A (t) / 2N = f A (t)
3 9/16/14 Genetic Drift
v what happens when we relax the
assumption of infinite population size?
² allele frequencies change due to random
“genetic drift” (stochastic sampling)
Genetic Drift
v Binomial probability
Pr{ j alleles of type A} =
€
! 2N $
(2N )! f j f 2 N− j
##
&& f Aj fa2 N− j =
A a
j
j!
2N
−
j
!
(
)
"
%
4 9/16/14 Random samples of 16 alleles for fA = fa = 0.50
frequency (proportion
of populations)
0.20
0.15
0.10
0.05
0.00
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16
number of A alleles
Random samples of 16 for fA = 0.25, fa = 0.75
frequency (proportion
of populations)
0.25
0.20
0.15
0.10
0.05
0.00
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16
number of A alleles
5 9/16/14 The Moran model
v derived for a haploid population
² but results for 2N haploid individuals are
essentially the same as for N diploid
individuals when assuming neutrality
v randomly select an adult to produce 1
offspring, then randomly select an adult
to die
v results in allele frequency increments of
±1
The Wright-Fisher model
v transition probabilities obtained directly
from binomial probability distribution with
fA and fa adjusted each generation
! 2N $! i $ j ! 2N − i $2 N− j
(2N )! f j f 2 N− j
&&#
Tij = ##
=
& #
&
A a
j!( 2N − j )!
" j %" 2N % " 2N %
v variance in allele frequency change for
a single generation = fAfa/2N
² (= 2fAfa/2N for 2N iterations of the Moran process)
6 9/16/14 Wright-Fisher transition probabilities
v binomial results in a matrix of transition
probabilities that can be multiplied to predict
the distribution of allele frequencies at any
number of generations in the future
to
matrix for
2N = 8
from
i
0
1
2
3
4
5
6
7
8
0
1
0.344
0.100
0.023
0.004
~0
~0
~0
0
1
0
0.393
0.267
0.112
0.031
0.005
~0
~0
0
2
0
0.196
0.311
0.235
0.109
0.030
0.004
~0
0
3
0
0.056
0.208
0.282
0.219
0.101
0.023
0.001
0
j
4
0
0.010
0.087
0.211
0.273
0.211
0.087
0.010
0
5
0
0.001
0.023
0.101
0.219
0.282
0.208
0.056
0
6
0
~0
0.004
0.030
0.109
0.235
0.311
0.196
0
7
0
~0
~0
0.005
0.031
0.112
0.267
0.393
0
8
0
~0
~0
~0
0.004
0.023
0.100
0.344
1
P = 100 populations
number of populations
25
20
15
10
5
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
number of A alleles
7 9/16/14 Predicted distributions of allele frequencies in
replicate populations of N = 16
Random genetic drift in 107 Drosophila populations;
8 males, 8 females each generation
data from Buri (1956)
8 9/16/14 Questions
v How does time to fixation (or probability
of fixation within a given number of
generations) depend on population
size?
v How does the probability of fixation for a
given allele depend on its initial
frequency?
v How likely is a new mutation to drift to
fixation?
Models of Genetic Drift
v Binomial Probability Distribution
v Markov Chains
v Diffusion Approximation
² based on continuous rather than discrete
variables (i.e., time and allele frequencies)
9 9/16/14 Wright-Fisher transition probabilities
v binomial results in a matrix of transition
probabilities that can be multiplied to predict
the distribution of allele frequencies at any
number of generations in the future
to
matrix for
2N = 8
from
i
0
1
2
3
4
5
6
7
8
0
1
0.344
0.100
0.023
0.004
~0
~0
~0
0
1
0
0.393
0.267
0.112
0.031
0.005
~0
~0
0
2
0
0.196
0.311
0.235
0.109
0.030
0.004
~0
0
3
0
0.056
0.208
0.282
0.219
0.101
0.023
0.001
0
j
4
0
0.010
0.087
0.211
0.273
0.211
0.087
0.010
0
5
0
0.001
0.023
0.101
0.219
0.282
0.208
0.056
0
Predicted distributions of allele frequencies in
replicate populations of N = 16
Pi→ j =
6
0
~0
0.004
0.030
0.109
0.235
0.311
0.196
0
7
0
~0
~0
0.005
0.031
0.112
0.267
0.393
0
8
0
~0
~0
~0
0.004
0.023
0.100
0.344
1
(2N )! f j f 2 N− j
A a
j!( 2N − j )!
10 9/16/14 The Diffusion Approximation
φ ( p, x;t + Δt ) − φ ( p, x;t ) =
−[ M ( x )φ ( p, x;t ) − M ( x − Δx )φ ( p, x − Δx;t )]
+
1
{[V ( x + Δx )φ ( p, x + Δx;t) − V ( x )φ ( p, x;t )]
2
−[V ( x )φ ( p, x;t ) − V ( x − Δx )φ ( p, x − Δx;t )]}
€
v “the discussion may seem quite mysterious, but
please do not be discouraged because a
detailed understanding is not necessary to
understand the rest of the chapter or anything
later in the book”
11 9/16/14 Expectations based on the diffusion approximation
Insights…
v time to fixation and time to loss
Tfix = −4N
(1− p) ln (1− p)
Tloss = −4N
p
p ln ( p)
1− p
v persistence time (for polymorphism)
Tsegregate = −4N "# p ln ( p) + (1− p) ln (1− p)$%
12 9/16/14 Insights…
v probability of fixation for a new,
neutral mutation = 1/(2N)
²  time to fixation = ~ 4N
v probability of loss = 1 - (1/(2N))
²  time to loss = ~ 2 ln(2N)
* these are average times!
13 9/16/14 100000"
100000"
2*ln(2N)"4"('me"to"loss)"
10000"
Number'of'Genera-ons'
Number'of'Genera-ons'
4N"('me"to"fixa'on)"
1000"
100"
10"
10000"
1000"
4N"(-me"to"fixa-on)"
2*ln(2N)"9"(-me"to"loss)"
100"
10"
1"
1"
1"
10"
100"
Popula-on'Size'
1000"
0"
1000"
2000"
3000"
4000"
5000"
6000"
7000"
8000"
Popula-on'Size'
14 
Download