National Technical University of Athens (NTUA)
Ανάπτυξη υβριδικών αλγορίθμων με βάση μεθοδολογίες δικτύων για τη διερεύνηση συσχετίσεων σε βιολογικά/περιβαλλοντικά δεδομένα
Abstract
dc:descriptionIn the present PhD Thesis, network-based algorithms aiming at analyzing and revealing interrelations within biological/environmental data are proposed. The algorithms are applied here within two different contexts: i) analysis of protein sequence and structural space, and fold recognition using networks as simple graphs (first part) and ii) analysis of multifactorial disease-related traits using hybrid artificial neural network methods (second part). In the first part, protein similarity networks are constructed separately for two similarity criteria based on sequence derived features and structural alignment. Measurements, like network degree, clustering coefficient, characteristic path length and vertex centrality are utilized to characterize their overall and local topology. Protein similarity networks are classified as small world networks, an architecture that can host the similarity transition among proteins during evolution. Furthermore, the task of fold recognition on a protein similarity network basis is addressed. A classifier that recognizes the fold of a query protein sequence by calculating its affinities to sequences-vertices of the protein sequence network is, thus, constructed and compared with popular classifiers. The classifier is then optimized by employing a genetic algorithm which selects the subset of sequence-derived features yielding a protein sequence network that is most similar to the corresponding protein structure network. In the second part, two artificial neural network based methods have been designed and developed towards the analysis of multifactorial traits that have been related to cardiovascular diseases. The proposed methods combine an artificial neural network with either a parameter decreasing algorithm or a genetic algorithm that optimizes the training process and the input features set fed to the network. Both methods were used to determine the most important factors (among sex, age, genetic polymorphisms, clinical measurements and measurements that describe average nutrition habits) that affect postprandial lipemia and human weight. The artificial neural network based methods were successful in classifying subjects into output classes derived from the multifactorial traits.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2009
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Valavanis, Ioannis
- Βαλαβάνης, Ιωάννης
Subjects
dc:subject × 36- Δίκτυα
- Γράφος
- Τεχνητά νευρωνικά δίκτυα
- Γενετικοί αλγόριθμοι
- Ταξινομητές
- Πρωτεϊνική ακολουθία - δομή
- Δίκτυο μικρού κόσμου
- Κεντρικότητα διαμεσότητας
- Αρχαιότητα πρωτεϊνών
- Πολυπαραγοντικός γενετικός πολυμορφισμός
- Παχυσαρκία
- Μεταγευματική λιπαιμία
- Γενετικός πολυμορφισμός
- Κύριος κόμβος
- Αναγνώριση διπλώματος
- Καρδιαγγειακές νόσοι
- Διατροφή
- Networks
- Graph
- Artificial neural networks
- Genetic algorithm
- Protein sequence - structure
- Small world network
- Betweenness centrality
- Protein ancestrality
- Genetic polymorphisms
- Obesity
- Classifier
- Hub
- Fold recognition
- Cardiovascular diseases
- Postprandial lipemia
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/18615