Αλγόριθμοι διαχείρισης και ανάλυσης ακολουθιών βιολογικών δεδομένων με εφαρμογή σε προβλήματα βιοπληροφορικής

Abstract

The object of this doctoral thesis is the study and the design of efficient algorithms for the analysis of sequences of biological data. The algorithms that we describe have application on Bioinformatics problems, such as the recognition of known or unknown patterns in DNA and RNA that are involved in various biological activities, as well as the discovery of periodicities. More specifically the algorithms that we present are used for the analysis of Biological Sequences with “don't care characters”' and Weighted Biological Sequences. Biological Sequences with “don't care characters”, usually represent protein families while Weighted Biological Sequences represent assembled sequences of genomes that they have been recently sequenced. In Biological Sequences with “don't care characters”' we present two efficient algorithms of linear time for the computation of the period and the cover. The second algorithm is also applied in circular DNAs . In Weighted Biological Sequences we present two algorithms for the computation of basic periodicities: the period and the cover, while we also solve the problem of pattern matching. The need for efficient management of biological sequences with weights prompted us to introduce a new efficient data structure which solves efficiently the two precedents problems. This structure is named Weighted Suffix Tree. Using the Weighted Suffix Tree we solve various instances of the motif discovery problem in Biological Weighted Sequences. Finally we decided to study the use of Genetic Algorithms and Evolutionary Programming in the analysis of biological sequences. The result of this study is the description of a genetic algorithm that computes the repetitions in a biological sequence.