Benjamin Good

1
The Optimality of the Genetic Code

• Benjamin Good
• March 17, 2008

2
The Genetic Code
? Codon

• Sequence constructed from 4 letters known as
  nucleotides or bases, denoted A, G, C, U
  / T
• These letters form fixed length words known as
  codons.
• Groups of codons form sentences which encode
  proteins.

3
The Genetic Code

• A given codon can either stand for a specific
  amino acid or act as a start/stop codon, which
  signals either the beginning or end of a
  proteins code respectively.
• There are 44464 different codons but only 20
  amino acids to code for, making a total of 21
  different possible meanings for a given codon
  (including start/stop).
• How are codons distributed among the 21 different
  categories?

4
The Genetic Code

• The Canonical Code
• But why this arrangement and not another?
• Crick Canonical code is a frozen artifact of a
  code that was good enough to work

5
Why the canonical code?

• An alternative is that the canonical code itself
  evolved to optimize for some selected trait.
• Noting the connection between similar codons and
  similar amino acids, several researchers
  hypothesized that the canonical code evolved to
  optimize against copying/transcription errors.

6
The Polar Requirement

• Woese and Alf-Steinberger came up with a measure
  for error susceptibility in genetic code based on
  hydrophobicity.
• A given codon is subject to a single mutation.
  The polar difference between the new amino acid
  and the old one is calculated.
• The error resulting from the mutation is taken
  as the distance squared (mean squared distance).

7
How Optimal is the Canonical Code?

• Unfortunately, Alf-Steinbergers results have not
  been reproducible.
• The first reproducible test of the polar
  requirement was published by Haig and Hurst in
  1991.
• Using this method, they calculated the total
  error for a large sample of possible code
  assignments.
• Out of 10,000, only twoother codes had
  lowererror values than thecanonical code!

8
One in a million?

• Freeland and Hurst built upon HHs model to
  introduce more realistic assumptions.
• Two types of code errors possible transition
  andtransversion.
• Introduced weighting fortwo types of errors
  because they are not equally probable in
  nature.
• Also introduced bias towardsmistranslation
  rather thanmutation (higher rates oferrors in
  1st and 3rd slots)

9
One in a million?
Weighted errors make the canonical code even more
optimized relativeto the rest.
Peak efficiency Around w 3
10
One in a million?

• Out of a sample
• of 1,000,000
• random codes,
• only 1 had a
• lower error value
• than the CC!
• It was relatively
• far away in
• search space,
• but behaved
• similarly to CC.

11
Beyond the Polar Requirement

• In the paper we read for class, Freeland and
  Hurst question previous studies (including their
  own).
• Is the polar requirement a biased measurement?
• Is using the (W)MSD a biased measurement?
• Some biosynthetic acids might be tied to
  particular codons, so code space could be
  artificially symmetric.
• Proposed a new measurement based on PAM matrices,
  which measure the similarity of two amino acids
  on a functional level.

12
Beyond the Polar Requirement

• General error metric

ei is the physical error resulting from
substitution i
A codes total error
ai is the number oftransition errors leadingto
substitution i.i.e. U ? C,A?G
PAM matrix
ßi is the number oftransversion errors
leadingto substitution i.i.e. U,C ? A,G
Polar requirement
13
Beyond the Polar Requirement

• Results

PAM Matrix
Polar Requirement
Far from overturning the adaptive hypothesis,
this new study showed the canonical code to be
even more optimized than previously thought!
14
Other optimizations

• Studies of the assignment of stop codons found
  that the canonical code is highly optimized
  against frameshift and nonsense mutations. (S.
  Naumenko et al., 2008)
• Furthermore, these same optimizations against
  frame shift errors allow the CC to be more
  efficient at encoding parallel information on top
  of a protein coding sequence. (Itzkovitz and
  Alon, 2007)

15
Is the canonical code optimized?

• YES!
• But many aspects are still unclear e.g. a
  mechanism for code selection.
• Conditions in precanonical times are still
  relatively unknown and the canonical code seems
  to be universally adhered to in modern organisms.

16
The End