Υπολογιστική μέθοδος σύγκρισης φαρμακολογικών ουσιών μέσω γονιδιακής έκφρασης στο μεταγραφικό επίπεδο

Σάρδης, Αντώνιος; Sardis, Antonios

dc.contributor.author	Σάρδης, Αντώνιος	el
dc.contributor.author	Sardis, Antonios	en
dc.date.accessioned	2020-02-03T11:49:02Z
dc.date.available	2020-02-03T11:49:02Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/49743
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.17441
dc.rights	Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα	*
dc.rights.uri	http://creativecommons.org/licenses/by-nc-nd/3.0/gr/	*
dc.subject	Φάρμακο	el
dc.subject	Μηχανική μάθηση	el
dc.subject	Γονίδια	el
dc.subject	Βιολογική απόσταση	el
dc.subject	Σύγκριση ουσιών	el
dc.subject	Machine learning	en
dc.subject	Drug discovery	en
dc.subject	Biological distance	en
dc.subject	Gene expression	en
dc.subject	Drug repositioning	en
dc.title	Υπολογιστική μέθοδος σύγκρισης φαρμακολογικών ουσιών μέσω γονιδιακής έκφρασης στο μεταγραφικό επίπεδο	el
dc.title	A computational method for comparing pharmacological compounds based on gene expression	en
heal.type	bachelorThesis
heal.classification	Βιοπληροφορική	el
heal.language	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2019-10-02
heal.abstract	Compound identification is the most crucial step in the drug discovery process. Today several computational methods focusing on chemical structure that aim to discover the optimal structure for a specific target or disease exist. Since the advent of the Connectivity Map and the release of large scale gene expression data following compound treatment, new methods that aim to discover compounds based on their biological function similarity have been proposed. These methods aim to discover compounds that cause a specific effect on cellular models, rather than relying purely on their chemical structure. The consensus of the two avenues (structure-based and function-based) is that similar compounds in structure will cause similar effects, but the opposite rarely holds. The aim of this project is twofold. First, to create new methods of calculating biological effect similarity based on gene expression data from compound perturbations. Second, to create a model that can predict the biological effect similarity of compounds from structural data, thus augment traditional structure-based screening approaches. On this front, first of all, the only attempt made on the above mentioned purposes is recreated and examined. It is about investigating the correlation between the chemical structure and the gene expression similarity of compound pairs. Afterwards, novel ways of processing the data at the transcriptional level are assessed and compared, aiming to the discovery of a characteristic able to decide whether a pair of pharmaceutical compounds are similar or not. Moreover, a new method of comparing compounds is defined and investigated. The foundation of this method consists of prior biological knowledge about protein networks and is then compared to the already existing ones. Finally, a machine learning model, which utilizes structural data in order to determine the pairwise transcriptional level similarity of compounds, is tested.	en
heal.advisorName	Αλεξόπουλος, Λεωνίδας
heal.committeeMemberName	Αντωνιάδης, Ιωάννης
heal.committeeMemberName	Κυριακόπουλος, Κωνσταντίνος
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Μηχανολόγων Μηχανικών. Τομέας Μηχανολογικών Κατασκευών και Αυτομάτου Ελέγχου	el
heal.academicPublisherID	ntua
heal.numberOfPages	75 σ.
heal.fullTextAvailability	false