Συντάχθηκε 15-10-2013 14:00
από Balasi Panagiota
Email συντάκτη:
Ενημερώθηκε:
-
Ιδιότητα: -.
ΠΟΛΥΤΕΧΝΕΙΟ ΚΡΗΤΗΣ
Σχολή Ηλεκτρονικών Μηχανικών και Μηχανικών Υπολογιστών
Πρόγραμμα Προπτυχιακών Σπουδών
ΠΑΡΟΥΣΙΑΣΗ ΔΙΠΛΩΜΑΤΙΚΗΣ ΕΡΓΑΣΙΑΣ
ΟΡΕΣΤΗ ΞΕΝΗ
με θέμα
Study, analysis and implementation of gene alignment algorithm on a platform based on reconfigurable hardware
Τετάρτη 16 Οκτωβρίου 2013, 1μμ
Αίθουσα 2042, Κτίριο Επιστημών, Πολυτεχνειούπολη
Εξεταστική Επιτροπή
Καθηγητής Απόστολος Δόλλας (επιβλέπων)
Καθηγητής Διονύσιος Πνευματικάτος
Αναπληρωτής Καθηγητής Ιωάννης Παπαευσταθίου
Abstract
Computational Biology is one of the evolutionary scientific areas that the Electronics and Computer Engineers study. The Bioinformatics results can be used in biology, medicine and pharmaceutics and they can lead into new medicines and therapy methods. The Bioinformatics area consists of really high compute intensive and resource demanding problems.
The Bowtie algorithm is a sequence alignment algorithm that was introduced in 2009. This algorithm uses a different methodology and techniques from the other sequent alignment algorithms. In more details, the Bowtie algorithm uses the Burrows & Wheeler compression method and the FM – Index technique, which is a pattern matching method for very fast search for similar patterns. The Bowtie algorithm takes as input an organism’s genetic database and genetic sequences-“queries” with high or low similarity to the input database. The algorithm’s result shows the similarity between each input query and the genetic database.
This thesis presents a reconfigurable hardware-based implementation of the Bowtie algorithm. The search process of the algorithm, which is the most compute intensive part, is implemented on a multi-FPGA platform. The final system offers one order of magnitude execution speedup vs. the official software, as far as the most time consuming part of the Bowtie algorithm.
