Inference of selection pressure

1
Inference of selection pressure
2
rate of non-synonymous substitution gtgt rate of
synonymous substitution gt positive
selection i.e. dN gtgt dS i.e. number of
nonsynonymous substitutions per nonsynonymous
site gtgt number of synonymous substitutions per
synonymous site
3
Calculation of dN and dS require quite complex
models of nucleotide substitution
Leu CTG
Leu TTG
Arg CGA
Trp TGG
4
It is rare that an entire gene is evolving under
positive selection Comparison of a pair of
sequences normally results in dN lt dS even if
some sites in the sequence are evolving under
selective pressure because many sites are
constrained Use site-specific methods
5
Adaptsite (Suzuki and Gojobori) - count the
number of nonsynonymous and synonymous
substitutions at a specific site - perform
statistical test for each site - consider
multiple testing
6
Codeml (Yang) - uses codon substitution model -
places individual sites in site classes -
allows likelihood ratio test to infer whether
there is evidence for some sites under positive
selection
7
Codeml Example The neutral model Two site
classes dN/dS 0 dN/dS 1 The selection
model Three site classes dN/dS 0 dN/dS
1 dN/dS free Some sites can be said to be
evolving under ve selection if - additional
parameter is justified AND - estimated (free)
value of dN/dS gtgt 1
8
General Comments - both codeml and Adaptsite
require a phylogenetic tree - Adaptsite is far
more conservative than codeml - recent research
has shown that recombination can give a false
positive signal