Genetic Distance

1
Genetic Distance
2
Definition A measure of how different two
sequences are The number of evolutionary events
that have occurred since two sequences diverged
Simplest p-distance proportion of sites
that are different (but p-distance is not a good
metric doesnt add ?)
3
A T T G C G C
C
T
C
A
A T T G C G C
C
T
A
A
T
A
Differences
Substitutions
4
Correcting for multiple substitutions
Jukes and Cantor all substitutions were created
equal
Kimura 2-parameter distinguish between
transitions and transversions
Models of nucleotide substitution correct for 2
things - biases in the rate at which different
mutations occur - biases in the frequency at
which different nucleotides occur
5
Jukes and Cantor d -3/4 ln (1 4/3p)
6
Genetic Distance
Proportion of sites that are different
7
Kimura 2P model Allows for transition-transversi
on bias HKY model Allows for nucleotide bias and
transition - transversion bias General
Reversible Allows for nucleotide bias as well as
different mutation rates for all nucleotide pairs
8
Jukes-Cantor
9
Kimura 2P
10
HKY
11
Note Be careful not to confuse the
transition/transversion rate ratio (k) with the
transition/transversion ratio (R)
In the case of the Kimura 2-P model
12
Additional corrections for different rates of
evolution at different sequence positions
can also be applied - gamma distribution
(incorporated analytically) - by codon position
13
In order to perform a gamma correction for site
specific rates you need to know the shape of the
gamma distribution
14
You need several sequences to work out the shape
of the gamma distribution You normally dont
apply a gamma correction on just a pair of
sequences unless you know the shape (alpha) of
the gamma distribution in advance
15
Codon substitution model
16
Some general points - genetic distances can be
far greater than 1 - smaller genetic distances
are more reliable - model choice has a bigger
impact for distantly related sequences -
normally positions with gaps are ignored
(complete deletion) - IF you know the rate of
evolution for a pair of sequences (and if the
rate has remained more or less constant) you can
estimate the date at which they diverged