Exon prediction by Genomic Sequence alignment Burkhard Morgenstern and Oliver Rinner

1

Burkhard Morgenstern Institut für Mikrobiologie
und Genetik Molekulare Evolution und
Rekonstruktion von phylogenetischen Bäumen WS
2006/2007


2
Tools for multiple sequence alignment

• s1 - R Y I - M R E A Q Y E S A Q
• s2 - R C I V M R E A - Y E - - -
• s3 - Y - I - M Q E V Q Q E R - -
• s4 W R Y I A M R E - Q Y E - - -
• Alignment hypothesis about sequence evolution
• Search for most plausible scenario!
• Estimate probabilities for individual
  evolutionary events insertions/deletions,
  substitutions

3
Substitution matrices

• All protein alignment programs depend on
  similarity scores s(a,b)
• Similarity score s(a,b) for amino acids a and b
  is based on probability pa,b of substitution a
  -gt b
• Idea it is more reasonable to align amino acids
  that are often replaced by each other!

4
Tools for multiple sequence alignment
Substitution matrix assigns score s(a,b) to every
pair of amino acids a and b
5
Substitution matrices

• Compute similarity score s(a,b) for amino acids a
  and b
• Probability pa,b of substitution
• a ? b (or b ? a),
• Frequency qa of a
• Define
• s(a,b) log (pa,b / qa qb)

6
Substitution matrices

• Simplifying assumptions
• Evolution as random process substitution a -gt b
  occurs with probability pa,b(t) depending on
  time t in evolution since sequences originated
  from common ancester
• pa,b(t) does not depend on sequence position
• Sequence positions independent of each other
• pa,b(t) pb,a(t) (symmetry!)

7
Substitution matrices

• For small values of t pa,b(t) grows linearly
  with time t for a ? b
• For large values of t multiple mutations must be
  accounted for

8
Substitution matrices

• To calculate pa,b
• Consider alignments of closely related proteins
  and count substitutions
• a ? b (or b ? a)

9
Substitution matrices

• To calculate pa,b
• Consider alignments of closely related proteins
  and count substitutions
• a ? b (or b ? a)
• ESWTS-RQWERYTIALMSDQRREVLYWIALY
• ERWTSERQWERYTLALMS-QRREALYWIALY

10
Substitution matrices

• To calculate pa,b
• Consider alignments of closely related proteins
  and count substitutions
• a ? b (or b ? a)
• ESWTS-RQWERYTIALMSDQRREVLYWIALY
• ERWTSERQWERYTLALMS-QRREALYWIALY

11
Substitution matrices

• Calculation of pa,b(t)
• Consider multiple alignments of closely related
  protein families
• Count substitutions a-gtb (or b-gta) in alignments
  based on phylogenetic tree
• Estimate pa,b(t) for small times t 1 PAM
  (Percentage of Accepted Mutations)
• Calculate conditional probabilities p(ab,t) for
  small t
• Calculate p(ab,t) for larger evolutionary
  distances by matrix multiplication
• Calculate pa,b(t) for larger evolutionary
  distances

12
Substitution matrices
13
Substitution matrices

• Alternative BLOSUM matrices
• S. Henikoff and J.G. Henikoff, PNAS, 1992
• Basis BLOCKS database, gap-free regions of
  multiple alignments.
• Cluster of sequences if percentage of similarity
  gt L
• Estimate pa,b(t) directly.
• Default values L 62, L 50