Grant: Master's Award: Canada Graduate Scholarships
McMaster University
$17,500.00
De-novo design of antibiotics targeting Acinetobacter baumannii using chemical language models
EF 2023-24
Résumé
The antimicrobial resistance (AMR) crisis is anticipated to result in 10 million deaths annually by 2050. Among the largest bacterial threats indicated by the World Health Organization is Acinetobacter baumannii, a nosocomial pathogen linked with over 300,000 deaths in 2019 alone. The threat of A. baumannii arises from its remarkable capacity to uptake extracellular DNA, facilitating the rapid acquisition of antibiotic resistance determinants. Combined with our limited antibiotic pipeline, this situation leaves us with few treatment alternatives moving forward. Amidst these challenges, Generative artificial intelligence (GAI) has emerged as an increasingly applied tool in de novo drug design and presents a promising avenue to design structurally and functionally novel antibiotics. Indeed, this project aims to develop a GAI algorithm for the design of antibiotic compounds targeting A. baumannii. We will leverage a class of GAI algorithms called chemical language models (CLMs) which generate string-based molecular representations (e.g. SMILES codes) and have shown utility in designing molecules with desired bioactivities in other domains. The CLM developed here, is trained with both antibacterial and non-antibacterial molecules—a novel approach in the CLM field—which significantly enriches antibiotic properties among generated molecules while simultaneously avoiding those that lack them, marking an improvement over prior CLMs. After generating several molecules with this CLM, they will be assessed for their predicted activity on A. baumannii and cytotoxicity utilizing a series of pre-trained machine learning property predictors. Subsequently, molecules with desirable physicochemical properties (e.g. solubility) will be chosen for in vitro and in vivo testing downstream.
Programme
Master's Award: Canada Graduate Scholarships
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database
Tel que publié
Id
505823
Name
Alexander Jeremie
Region
Ontario
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
De-novo design of antibiotics targeting Acinetobacter baumannii using chemical language models
Programtype2
Studentship Programs
Orgnameinp2
McMaster University
Cihramount2
$17,500
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Acinetobacter Baumannii; Antibiotic Resistance; Antibiotics; Artificial Intelligence; Chemistry; Drug Discovery; Language Models; Machine Learning
Abstract
The antimicrobial resistance (AMR) crisis is anticipated to result in 10 million deaths annually by 2050. Among the largest bacterial threats indicated by the World Health Organization is Acinetobacter baumannii, a nosocomial pathogen linked with over 300,000 deaths in 2019 alone. The threat of A. baumannii arises from its remarkable capacity to uptake extracellular DNA, facilitating the rapid acquisition of antibiotic resistance determinants. Combined with our limited antibiotic pipeline, this situation leaves us with few treatment alternatives moving forward. Amidst these challenges, Generative artificial intelligence (GAI) has emerged as an increasingly applied tool in de novo drug design and presents a promising avenue to design structurally and functionally novel antibiotics. Indeed, this project aims to develop a GAI algorithm for the design of antibiotic compounds targeting A. baumannii. We will leverage a class of GAI algorithms called chemical language models (CLMs) which generate string-based molecular representations (e.g. SMILES codes) and have shown utility in designing molecules with desired bioactivities in other domains. The CLM developed here, is trained with both antibacterial and non-antibacterial molecules—a novel approach in the CLM field—which significantly enriches antibiotic properties among generated molecules while simultaneously avoiding those that lack them, marking an improvement over prior CLMs. After generating several molecules with this CLM, they will be assessed for their predicted activity on A. baumannii and cytotoxicity utilizing a series of pre-trained machine learning property predictors. Subsequently, molecules with desirable physicochemical properties (e.g. solubility) will be chosen for in vitro and in vivo testing downstream.
Competitiondate
202312
Approvedterm2
1 yr 0 mth
Pinamesdelim
Alexander, Jeremie
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $17,500~ $0~