Analiza proteinskih poravnanja

master's thesis

• Title (english): Analysis of protein alignments
• Author: Eva Belcar
• Mentor: Pavle Goldstein (mentor)
• Committee member: Pavle Goldstein (predsjednik povjerenstva)
• Committee member: Goran Muić (član povjerenstva)
• Committee member: Tina Bosner (član povjerenstva)
• Committee member: Ozren Perše (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Mathematics); Zagreb
• Defense date and country: 2023-03-03, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Mathematics

Abstract

Tematika ovog rada je klasifikacija proteina u proteinsku familiju katepsina na temelju karakterističnih podnizova aminokiselina (motiva) i međusobnog uspoređivnja. Cilj je izbaciti što više lažno pozitivnih i pri tome zadržati stvarno pozitivne motive iz outputa pretraživača motiva. Problematika rada svodi se na prelazak u višedimenzionalni vektorski prostor prevođenjem aminokiselina u numeričke vektore koji predstavljaju određena obilježja tih aminokiselina. Pronalazi se kugla koja sadrži najviše motiva, uključujući i motiv katepsina. Prvo se računa benchmark za uspješnost modela, zatim se razvija model procjenjivanjem radijusa kugle i prolaskom po svim točkama za određivanje njezinog središta. Središte se pomiče u težište točaka u kugli i taj postupak se ponavlja do kada se središte ne stabilizira. Na kraju se model validira i uspoređuje s benchmarkom. U pet proteoma napravljena je klasifikacija katepsina. Rezultati pokazuju da model uspješno i brzo izbacuje lažne pozitivce, a da pritom zadrži većinu pravih pozitivaca neovisno o veličini početnog uzorka.

Abstract (english)

The subject of this thesis is classification of proteins into the Cathepsin protein family based on characteristic sequences of amino acids (motifs) and pairwise comparison. The goal is to reduce the number of false positives and keep true positive motifs from the output of a motif scanner. The problem comes down to transitioning into multidimensional vector space by interpreting amino acids as numeric vectors which represent certain characteristics of those amino acids. A ball containing most motifs is found, including the Cathepsin motif. Firstly, a benchmark for the performance of the model is calculated, then the model is being developed by estimating the expected radius and by optimizing the center of the ball. At the end, the model is being validated and the performance is compared to the benchmark. The classification of Cathepsins is made in five proteoms. The results show that the fast model successfully decreases the number of false positive motifs but manages to keep true positives regardless of the size of the initial sample.

• Keywords: klasifikacija proteina; proteinska familija katepsina; višedimenzionalni vektorski prostor; numerički vektori; benchmark za uspješnost modela
• Keywords (english): classification of proteins; Cathepsin protein family; multidimensional vector space; numeric vectors; benchmark for the performance of the model
• Language: croatian
• URN:NBN: urn:nbn:hr:217:726784
• Study programme: Title: Mathematical Statistics; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra matematike (magistar/magistra matematike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2023-03-23 11:55:14