Model dubokog učenja pore za sekvenciranje nanoporama

master's thesis

• Title (english): Deep Learning Model of Nanopore Sequencing Pore
• Author: Rafael Josip Penić
• Mentor: Mile Šikić (mentor)
• Committee member: Mile Šikić (predsjednik povjerenstva)
• Committee member: Marko Subašić (član povjerenstva)
• Committee member: Krešimir Križanović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-01, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Sekvenciranje nanoporom je jedna od najmodernijih metoda sekvenciranja bioloških sekvenci i u ovom radu smo istražili kako samo-nadzirajuće metode utječu na performanse modela zaduženih za određivanje baza i slične zadatke.

Isprobali smo metodu kontrastivnog prediktivnog kodiranja kojom smo pokušali predvidjeti buduće latentne reprezentacije signala. Nažalost, ovaj pristup nije uspio unaprijediti točnost modela.

Najbolje rezultate smo dobili metodama učenja kvalitetnih reprezentacija signala koje su iznimno popularne u području računalnog vida. Ključnu ulogu kod ovakih algoritama imaju transformacije signala i vrlo je važno smisliti dobre transformacije koje će pomoći modelu da nauči kvalitetne vektorske reprezentacije.

Abstract (english)

Nanopore sequencing is one of the most modern sequencing technologies and in this thesis we explored how do the self-supervised methods affect models that work with nanopore readings.

We tried out contrastive predictive coding method with which we tried to predict signal's future latent representations. Sadly, this approach did not improve model's performance on the downstream task.

We achieved best results with self-supervised representation learning methods which are very popular in the field of computer vision. Signal augmentations have a very important role in these algorithms and it is very important to think of good augmentations that will help model learn high-quality signal representations.

• Keywords: bioinformatika; samo-nadzirajuće učenje; sekvenciranje; nanopora; DNK; duboko učenje; strojno učenje
• Keywords (english): bioinformatics; self-supervised learning; sequencing; nanopore; DNA; deep learning; machine learning
• Language: english
• URN:NBN: urn:nbn:hr:168:713036
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 13:27:24