Back to search

University of British Columbia

Machine learning for antimicrobial peptide discovery and design

Abstract

dc:description

The growing global health concern of antibiotic resistance is prompting researchers to seek substitutes for conventional antibiotics. Antimicrobial peptides (AMPs), a diverse class of short and often cationic biological molecules, are gaining attention as promising candidates. While direct, large-scale wet lab screening is time-consuming and costly, using high-throughput bioinformatics tools to discover and design novel AMPs is an attractive approach. In this thesis, I introduce in silico tools for AMP discovery and design with machine learning models, and present the novel AMPs revealed by those methods. The in silico discovery of AMPs typically involves the investigation of huge genomics, transcriptomics, or protein datasets, and accurate methods to sift through such large volumes of candidate sequences are required. In this thesis, I introduce AMPlify, a deep learning based tool for AMP prediction, improving upon the state-of-the-art methods by incorporating attention mechanisms. By integrating AMPlify into bioinformatics pipelines or workflows, four novel AMPs with proven antimicrobial activity have been identified from the Rana [Lithobates] catesbeiana (bullfrog) genome, as well as 13 other novel AMPs mined from the UniProtKB/Swiss-Prot database. On the other hand, the potential sequence space of amino acids is combinatorially vast, allowing for the exploration of more AMPs that may not exist in nature to further expand the current arsenal of peptide-based therapeutics. However, manual design of novel synthetic AMPs requires prior field knowledge, restricting its throughput. In silico sequence generation methods for de novo AMP design stand out to be a high-throughput way to unearth novel synthetic AMPs. In this thesis, I introduce a recurrent neural network based tool, named AMPd-Up, for AMP sequence generation, and demonstrate its performance over existing methods. With AMPd-Up, 40 novel synthetic AMPs have been designed with proven antimicrobial activity against the bacterial strains tested in vitro. I demonstrate the utility of AMPlify and AMPd-Up in the discovery and design of novel AMPs, and I expect these tools to play an important role in our fight against antibiotic resistance.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy - PhD
Level thesis:degree_level
doctoral
Discipline thesis:degree_discipline
Bioinformatics
Grantor dc:publisher
University of British Columbia
Year dc:date
2024

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Li, Chenkai
Contributors dc:contributor
  • Birol, Inanc

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/87512

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Li, Chenkai. Machine learning for antimicrobial peptide discovery and design. doctoral thesis, University of British Columbia, 2024. http://hdl.handle.net/2429/87512