Correlated Mutations and Co-evolution

1
Correlated Mutations and Co-evolution

• May 1st, 2002

2
What is Co-evolution (Correlated Mutation)?

• Individual regions of proteins interact
• Regions can be either on the same chain or on
  different chains (complexes)
• A mutation in one half of the pair induces a
  change in the other half of the pair
• the tendency of positions in proteins to mutate
  co-ordinately Pazos et. al. 1997

3
Correlated Mutations Contain Information about
Protein-protein interactions Pazos et. al. 1997

• A possible aid to the docking problem, using
  only sequence information
• Docking The process by which protein domains
  interact with one another ? fitting

4
Methodology

• The correlation coefficient
• S is the similarity between residues at the
  positions i/j of type k versus l
• Arbitrarily chosen cutoff M predicted contacts
  (greatest L/2 values) i.e. ML/2

5

• The Harmonic Average (Xd)
• Measure of correlatedness
• Pic percentage of correlated pairs with that
  distance, Pia for all pairs

6
Comparisons of Correlations
7
Docking solutions test

• Note larger percentages imply worse performance
• Special mention of 2gcr and 3adk
• sequence information does not seem to be
  sufficient to discriminate

8
Figure 5 Scatter plot of Xd vs RMS
distance 9pap Hemoglobin 1hbb
9
Prediction Hsc70

• Figure 6 predicted contacts of Nt and Ct domains
  of Hsc70
• Could be verified experimentally

10
Coevolving Protein Residues Maximum Likelihood
and Relationship to Structure. Pollock et. al 1999

• Using size and charge characteristics to define
  co-evolution (correlation)
• Negative Correlation Correlation due to
  differences in charge (and thus also coevolution)

11
The Markov process model (simulated evolution)

• Two states, A and a
• Equation 1, the probability of transitioning
  state
• ? rate parameter
• p equilibrium frequency

12
Use of parameters in model

• Basic model for how they simulate evolutionary
  steps

13
Likelihood Test Characteristic (LR)

• LI and LD maximum likelihood values for
  independent and dependent model
• Method of determining whether dependence is
  statistically significant

14
Test of Significance (LR values for change in
parameters)
15
Myoglobin

• Used structure of myoglobin compared differences
  in sequences
• Variety of species used for sequence information
  sperm whale 3D protein structure

16
LR distributions for myoglobin size and charge

• Note the large negative correlation LR values in
  charge

17
Co-evolution of Proteins with their Interaction
Partners, Goh et. al. 2000

• Applied to PGK
• Chemokines

18
What is PGK?
19
Methodology

• Two independent sequence alignments, for N and C
  regions, using PSI-BLAST
• ClustalW to create distance matrix between
  complete domains
• To determine correlation, used equation below
• X and Y correspond to domains r a measure of
  relatedness between these domains

20
PGK correlations
21
Chemokines

• Role of chemokines importance in immunity (HIV,
  cancer)
• Four categories, mean nothing to me

22
Clustering of Chemokines
23
Clustering of Chemokine receptors