Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

Protein Structure Alignment using a Genetic algorithm by fop21123

VIEWS: 13 PAGES: 11

									     Protein Structure Alignment
     using a Genetic algorithm
                 By Szustakowski et al
Proteins:Structure, Function, and Genetics(38):428-440,2000
                Presented by Nannan Li



                                                              1
Introduction
   To establish evolutionary relationships
    between the proteins




                                              2
Biological problem
   For many protein pairs, distinct
    alignments could be generated that are
    indistinguishable in terms of number of
    equivalent residues and root mean
    square error of superposition
   Protein structures are more conserved
    in the core than in exposed loops and
    turns
                                              3
Motivation
   To develop a structure alignment
    algorithm with the goal of generating
    high-quality, biologically meaningful
    alignments by first aligning the protein‟s
    cores (secondary structure elements)




                                             4
Method
   Target Function--resulting in correct pairing
    of SSEs. “Elastic similarity score” has adopted
    to simultaneously maximize the number of
    equivalent residue pairs and minimize the
    distance between these pairs
   Treating each protein as a collection SSEs to
    avoid exhaustive search for regions of
    similarity shared by two distance matrices


                                                  5
Genetic Algorithm
   Use genetic algorithm to search optimal
    solution to target function
   Algorithm starts from a population of
    completely random pairs of alignment and
    happens in generations. Multiple SSE
    alignment are stochastically selected from
    the current population, modified (mutated or
    recombined) to from a new population,
    which becomes current in the next iteration
    of the algorithm
                                               6
Genetic Algorithm Steps
1)   Generate an initial population for possible SSE
     alignments
2)   Alter each alignment using “mutate"," hop”, and
     “swap” operators
3)   Carry out “recombination” between randomly
     assigned pairs of alignments using the “crossover”
     operator
4)   Accept or reject the alterations made to each
     alignment
5)   Exit if certain conditions are met. Otherwise go to
     step 2

                                                           7
Initial Population
   Since SSE alignment search space is
    very large, we biased the initial
    population toward SSE pair doublets
   Similarity scores are then calculated for
    all SSE pair doublets based on target
    function (Population size is set to 100)


                                                8
Genetic algorithm operators
   “mutate”– with a mutation probability,
    mutate the individual SSE pairs at each
    residue pairs
   “hop”– with a hop probability, two SSE
    pairs in one selected alignment trade
    places
   “swap”– with equal probability, an
    alignment is swapped with its parter

                                              9
Genetic algorithm
operator(„contd)
   Crossover– each alignment is randomly
    assigned a crossover partner from the
    rest of the population




                                        10
Availability

   C++ program called KENOBI
    http://zlab.bu.edu/k2/documents.shtml




                                        11

								
To top