Određivanje bakterijskih gena rezistentnih na antibiotike

undergraduate thesis

• Title (english): Identification of bacterial antibiotic resistance genes
• Author: Marko Žagar
• Mentor: Mirjana Domazet-Lošo (mentor)
• Committee member: Mirjana Domazet-Lošo (predsjednik povjerenstva)
• Committee member: Mihael Kovač (član povjerenstva)
• Committee member: Andrej Slapničar (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-06-27, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Rezistencija na antibiotike je aktualan problem u medicini. Problem je ozbiljan zbog
horizontalnog prijenosa gena koji brzo širi rezistenciju između bakterija. Jedan od
načina na koji se može identificirati i klasificirati rezistencija su tehnologije umjetne
inteligencije poput jezičnih modela i strojnog učenja.
U ovom radu sam iskoristio skup javno dostupnih sljedova gena koji stvaraju
rezistenciju iz CARD baze podataka te sam s pomoću ESM-1b proteinskog jezičnog
modela stvorio numeričke... More reprezentacije tih gena. Zatim sam trenirao dva XGBoost
modela strojnog učenja. Jednog sam trenirao da identificira, a drugog da klasificira
gene koji stvaraju rezistenciju.
Model može s točnosti od 97 % identificirati uzrokuje li neki gen rezistenciju, a ako
gen stvara rezistenciju model može s točnosti od 97 % odrediti na koji je antibiotik
gen rezistentan. Takav model bi mogao naći mjesto u praksi kao pomoć prilikom
odabira tretmana. Less

Abstract (english)

Antibiotic resistance is a current problem in medicine. One of the ways to identify
and classify resistance is through artificial intelligence technologies such as language
models and machine learning.
In this study, I utilized a publicly available dataset of resistance gene sequences from
the CARD database and used a protein language model to create numerical
representations of these genes. I trained two XGBoost machine learning models. One
model was trained to identify, and the other... More to classify genes that confer resistance.
The trained model can identify with 97 % accuracy whether a gene causes resistance,
and if the gene causes resistance, the model can determine with 97 % accuracy to
which antibiotic the gene is resistant. Such a model could find a place in practice as
an aid in selecting treatment. Less

• Keywords: bioinformatika; rezistencija na antibiotike; strojno učenje; transformer; duboko učenje; XGBoost
• Keywords (english): bioinformatics; antibiotic resistance; machine learning; deep learning; XGBoost
• Language: croatian
• URN:NBN: urn:nbn:hr:168:791668
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-10-12 22:34:31