L28 Human Population Genetics 1

Question 1

How many ancestors do you have?

Answer 1

see onenote slides

Can one person be multiple ancestors at once?

Question 2

How many ancestors are you genetically related to?

Answer 2

see onenote

How much did each of them contribute to your genome?

From grandparents we don’t get a clean cut 25% of their genome from them, can be around 23% or 27%
After parents, there is uncertainty of how much of our ancestor’s genome we inherit (inherit 50% from mum and 50% from dad)

Question 3

Humans are not infinite

Answer 3

see onenote

ancestors at generation
n=2^n

Question 4

Who does your genome come from?

Answer 4

see onenote slides

% genome inherited from each

parent: 50%
gparent: 25%
ggparent: 12.5%
gggparent: 6.75%

Question 5

Human population genetics

Answer 5

All humans are related due to our evolutionary history

but the specifics are trickier:

1. when?
2. where?
3. how much?
4. what does it mean for you?
5. what does it mean for the species?

population genetics: the extension of Mendelian genetics to evolving populations

Question 6

How do we talk about pop gen?

Answer 6

see onenote

need a formal means to describe processes such as:
- heterozyosity
- relatedness
- diversity
- mutations
- etc.
to synthesise these observations into coherent knowledge

Question 7

Multiple ways of thinking about a population

Answer 7

see onenote

1. full genealogy
2. ancestry of extant lineages
3. ancestry of sampled lineages
4. coalescent of sampled lineages

Question 8

Our most basic tool - a single segregating locus

Answer 8

see onenote

single diploid locus
2 alleles: A and a
p = A freq.
q = a freq.
p + q = 1

Question 9

Hardy-Weinberg equilibrium

Answer 9

see onenote

relationship between genotype and allele frequencies knows as h-w equilibrium

for it to hold, multiple assumptions must be true:

1. random mating
2. no selection
3. no migration
4. no mutation
5. no genetic drift (infinite population size)

Question 10

Beyond loci, to populations: the wright-fisher model

Answer 10

see onenote slides

still only one locus but now we explicitly model the passing of time

1. population size is constant (not infinite)
2. generations are discrete and non-overlapping
3. all individuals are equally fit
4. no mutation
5. haploid individuals
6. when extended to diploid mutations
   - non-assortative mating
   - no recombination

Wright-Fisher model, models genetic drift
- Binomial sampling with some extra parameters
Can be extended to incorporate things like selection

WF provides a robust null hypothesis for everything else we discuss

Question 11

Fundamental properties of the WF model

Answer 11

see onenote

each individual at generation t has 2N opportunities to become an ancestor to generation t+1

the probability of being ancestor to a specific individual is 1/2N

binomial distribution
n = 2N
p = 1/2N

Question 12

WF from the gene’s perspective

Answer 12

see onenote

there can be multiple copies of a gene in a generation

probability of being transmitted to generation t+1 now depends on number of copies of gene at t

still a random binomial process
n = 2N
p = x/2N

Question 13

Extending the model: WF with mutation

Answer 13

see onenote

introduce mutation into the population with probability “mu” per site per generation

in humans, “mu”~1 x 10^-8 per site per generation

Question 14

A window into the past: the coalescent

Answer 14

see onenote slides

can use assumptions of the WF model to go backwards and reconstruct the past from the present

Can answer questions such as:
- When did two individuals last share a common ancestor? When did they coalesce?
- Coalescent lets us date relatedness in a way that we can track
Can extrapolate current data back into the past to figure out relatedness between individuals

Question 15

Coalescent lets us date relatedness and divergence between sequences

Answer 15

see onenote

• given a DNA phylogeny, it allows us to find the time to most recent common ancestor (TMRCA) of sequences in the phylogeny
• coalescent modelling can be expanded to incorporate selection, expansion etc.
• gives us the framework against which to test predictions and observations

Question 16

But real populations are far more complicated

Answer 16

WF model is unrealistic but it gives us a mathematical language to describe expectations under neutrality, as well as mutation, selection, population growth etc.

does the parameter of the model resemble observed data?

WF and coalescent modelling empowers us to make robust judgements about actual populations and their genetic data

Question 17

Describing population structure

Answer 17

Population
- Randomly reproducing group of individuals

random mating = random relative to genetics
- meaning there is no underling population structure or subpopulations

but humans violate these assumptions:

1. geographic structure/isolation by distance
2. assortative mating
3. inbreedinge etc.

Question 18

F statistics for relationships between populations

Answer 18

see onenote

H statistics = measurements of heterozygosity and diversity

• Hi = heterozygosity within an individual
• Hs = heterozygosity within a subpopulation
• Ht = heterozygosity within the total population

F statistics = measure of population structure and differentiation

• Fxy = correlation between gametes drawn from x relative to y
• Fis = describes correlation (or diversity) between gametes from a single individual relative to what is seen in the subpopulation

F and H statistics are hierarchical

Fst is the most widely used F statistic in population genetics
- it describe the proportion of the variance due to between-population rather than within-population differences

Question 19

The international HapMap project

Answer 19

see onenote slides

aimed to generate a haplotype map of the human genome to be used in mapping genes associated with any disease

use of trios allowed for phasing of data and identification of haplotypes

Question 20

Haplotype

Answer 20

a block of SNPs physically on the same chromosome, inherited together due to LD

haplotypes are used to identify recombination break point, linkage blocks and tag SNPs

tag SNP = representation SNP for an entire haplotype

Question 21

HGDP-CEPH and 1000 genomes

Answer 21

see onenote

Human genome diversity project
- for ethnic and evolutionary studies

1000 genomes
- for broadly medical studies

Question 22

HGDP-CEPH - treasure trove or ethics quagmire?

Answer 22

see onenote

aim of the project was studies of human history and evolution - not medical research but all project data is publicly available so…

sought to collect DNA from isolated populations “at risk of becoming extinct” without working with the communities to prevent it

biocolonialism has resulted in many indigenous groups worldwide refusing to participate in genetic research when the distribution of benefits is inequitable