Refining the progressive multiple sequence alignment score using genetic algorithms


Ergezer H., Leblebicioglu K.

ARTIFICIAL INTELLIGENCE AND NEURAL NETWORKS, cilt.3949, ss.177-184, 2006 (SCI İndekslerine Giren Dergi) identifier

  • Cilt numarası: 3949
  • Basım Tarihi: 2006
  • Dergi Adı: ARTIFICIAL INTELLIGENCE AND NEURAL NETWORKS
  • Sayfa Sayıları: ss.177-184

Özet

Given a set of N (N > 2) sequences, the Multiple Sequence Alignment (MSA) problem is to align these N sequences, possibly with gaps, that bring out the best score due to a given scoring criterion between characters. Multiple sequence alignment is one of the basic tools for interpreting the information obtained from bioinformatics studies. Dynamic Programming (DP) gives the optimal alignment of the two sequences for the given scoring scheme. But, in the case of multiple sequence alignment it requires enormous time and space to obtain the optimal alignment. The time and space requirement increases exponentially with the number of sequences. There are two basic classes of solutions except the DP method: progressive methods and iterative methods. In this study, we try to refine the alignment score obtained by using the progressive method due to given scoring criterion by using an iterative method. As an iterative method genetic algorithm (GA) has been used. The sum-of-pairs (SP) scoring system is used as our target of optimization. There are fifteen operators defined to refine the alignment quality by combining and mutating the alignments in the alignment population. The results show that the novel operators, sliding-window, local-alignment, which have not been used up to now, increase the score of the progressive alignment by amount of % 2.