Simon Myers, Gil McVean

1
Recombination and genetic variation models and
inference

• Simon Myers, Gil McVean
• Department of Statistics, Oxford

2
What does the basic coalescent miss?

• The basic coalescent model is excellent at
  capturing the great stochasticity underlying
  genealogical processes in natural populations
• It does not, however, capture many biologically
  important features that we would like to learn
  about
• Recombination
• Changes in population size
• Geographical structure
• Natural selection
• We would like to extend the basic model to
  incorporate these features so that we can
• Look for their influence (model testing)
• Estimate their importance (parameter estimation)

3
Recombination at fine scales
Father
Mother
Child
Reciprocal cross-over
Gene conversion

• Region has two parental chromosomes
• Always one parent in the basic coalescent

4
Adding recombination

• Most genes in most genomes undergo recombination
• Reciprocal cross-over
• Gene conversion
• Transformation
• Template-switching
• For example, few gt200kb stretches of the human
  genome are free of recombination
• The effect of recombination on genealogies is
  described by the ancestral recombination graph
  (ARG)
• Today we will
• Examine the effects of recombination on patterns
  of genetic diversity
• Learn how to incorporate it in the coalescent
  model
• Look at how to learn about recombination from
  empirical data

5
Recombination influences how we think about gene
history
6
Ignoring recombination leads to problems in
inference

• Many inference procedures assume no recombination
• Presence of recombination can give misleading
  inferences

Mismatch distributions
Phylogenetic tree reconstruction
Schierup and Hein (2000)
7
Recombination is an important evolutionary
processes

• Recombination brings together multiple beneficial
  mutations
• Fisher/Muller
• Recombination brings together multiple
  deleterious mutations
• Mullers ratchet, Kondrashovs synergistic
  deleterious mutations
• Recombination allows beneficial mutations to
  escape from linked deleterious ones
• The ruby in the rubbish
• Recombination creates novel combinations of
  mutations
• Evasion of pathogens and parasites (Hamilton)
• Reduction in sib competition (Bell)

8
Recombination allows us to map genes by
association
Correlated, linked marker
Disease-associated mutation
Uncorrelated, unlinked marker
Much more about this on Thursday in week 2
(Jonathan Marchini)
9
Recombination gives us greater power in inference

• A single non-recombining locus gives us a single
  draw from the underlying genealogical process
• Looking at a single locus will only give us a
  small part of the populations history
• Recombination means that different loci have
  different draws from the genealogical process
  and therefore have independent information about
  processes we wish to learn about

10
Human history from the mtDNA and autosomal
perspectives

• Mitochondrial DNA is highly variable and
  non-recombining
• Phylogenetic trees of mtDNA have been used to
  make inferences about human history
• Out of Africa hypothesis with mtDNA eve about
  200,000 years ago
• Strong population growth
• But is it representative?
• Single evolutionary history
• Natural selection
• Small population size
• (Female demography only)
• Autosomal sequences
• Average TMRCA much older (1MY)
• Much less evidence for growth

Vigilant et al. (1991)
11
Practical what does recombination do to genetic
variation?

• Informally, recombination shuffles up genetic
  diversity
• We can see the effect of recombination in how
  structured genetic variation is

Chromosomes
Sites
Chromosome 22 1Mb 57 Europeans
Lipoprotein Lipase 10kb 48 African Americans
Xq13 10kb 69 worldwide