Efficient Algorithms for Triple-wise Alignment and Its Applications

Abstract

Sequence alignment is a scientific method that contributes to DNA homology studies, phylogeny determinations, and identification of conserved motifs. In the past few decades, pair-wise alignment has become a methodological standard used in many MSA methods. However, an increasing number of studies indicated that the three-way alignment, which is the alignment of three sequences, is able to provide additional information or a more accurate alignment result than what pair-wise alignment is able to give. In this dissertation, we focused on the investigation and application of three-way alignment algorithm. For the investigation of three-way alignment algorithm, we proposed two efficient methods, a dynamic programming-based algorithm with the variable gap penalty strategy and a linear algorithm adopting a probabilistic filtration model, to align protein and DNA sequences, respectively. For the application of three-way alignment, we applied three-way alignment to the methods and applications that originally adopt pair-wise alignment approaches. We presented a new progressive multiple sequence alignment strategy that combines pair-wise and three-way alignments to compare multiple sequences accurately. Similarly, we extended the three-way alignment algorithm to align three profiles to provide the different insight to the profile-profile alignment method. In addition, we developed a parallel algorithm for three-profile alignment to reduce the computational cost. Further, we combined the three-profile alignment approach and a voting algorithm to select the functional sites of the target protein by comparing protein superfamilies. Theoretical analysis and extensive experimental tests of the proposed methods are conducted in this dissertation. From the conducted experimental results, we got some encouraged remarks regarding to the proposed methods for sequence analysis.