A machine learning platform for drug repositioning against future infectious diseases
Grant Data
Project Title
A machine learning platform for drug repositioning against future infectious diseases
Co-Investigator(s)
Dr Song You-qiang
(Co-principal investigator)
Duration
24
Start Date
2023-06-01
Amount
150000
Conference Title
A machine learning platform for drug repositioning against future infectious diseases
Keywords
Antiviral drug discovery, Machine intelligence, Web application
Discipline
VirologyDatabase and data science
HKU Project Code
2203100624
Grant Type
Seed Fund for PI Research – Translational and Applied Research
Funding Year
2022
Status
On-going
Objectives
Key issues to be addressed:Globally, the community is unable to address rapidly evolving pandemic/epidemic agents in a timely manner by using the conventional one-bug-one-drug paradigm. Given the urgency and severity of a global pandemic, repositioning of existing drugs is the most practical route for the rapid identification of good candidates. The conventional high throughput screening exercise is resource-intensive and time-consuming. Instead, a machine learning networking that comprehensively merge structure-drug correlations, antiviral potency, pharmaceutical properties and safety profiles would enable rapid and precise identification of promising drug candidates for clinical trials against the emerging virus disease, once genomic information of an ‘unknown’ virus becomes available. Innovation of the project:Our proposal adapts broad fields of structural biology, artificial intelligence and drug development to this promising field of R&D. First, this suite of machine learning models can be used via the command line for large-scale virtual screening. As new datasets of drug bioactivity become available in the public domain or in house, we will tune the machine learning models for enhanced performance and external validation. In this regard, the platform enables ‘life-long’ machine learning. Second, the platform can decrease the number of experiments and reduce downstream costs. It helps the industry, scientific community and clinicians reduce the number of molecules before antiviral tests for any virus of interest. Finally, we will establish a set of unique model-to-website software online tools, which can be publically accessed by people in other fields beyond infectious diseases.