Scribe: N. Allen

Today's Topics

• Multiple Sequence Alignment
  • Typically uses proteins as alphabet
  • Start with: sequences s₁,s₂,...,s_k
  • Goal: produce alignment such as the ones on SM page 69 or 71
  • One system for scoring is sum-of-pairs score:
    • Each column scored seperately
    • Sum of pairwise scores of all pairs of symbols in the column
    • Dynamic programming implementation has time and space complexity n^k
  • Scoring heuristic:
    • Aggregate two strings using a dynamic programming based algorithm
    • Add additional strings to aggregate using a dynamic programming based algorithm
  • Alternative scoring system: Tree Alignment (SM page 79)
• Database Searching
  • Goal: approximate global alignment with a large database very fast
  • One approach uses PAM (% of accepted mutations) matrices
    • Works with number of mutations per 100 amino acid sequences
    • Only considers mutations that are positively selected by the environment (survivable mutation)
    • Considers mutations to be undirected events (A->Y equally likely as Y->A)
  • Computing a PAM matrix
    • Let p_e be the frequency of occurence for some amino acid e
    • Let f_ab be the number of a<->b mutations observed
    • Let f_a be the sum of f_ab for all b
    • Let f be the of f_a for all a
    • The matrix entries M_aa are then 1 - (f_a / (100*f*p_a))
    • The matrix entries M_ab are then (f_ab / f_a) * (f_a / (100*f*p_a))

Today's Sources

• Setubal and Meidanis: Chapter 3 (pages 69-82)