A breakthrough in computational drug discovery has emerged from Finland, where researchers have achieved a remarkable acceleration in virtual drug screening through advanced machine learning technology. The collaborative effort between the University of Eastern Finland (UEF) and Orion Pharma has demonstrated a dramatic reduction in screening time for potential drug candidates.
Revolutionary Speed in Drug Candidate Identification
The research team, led by Dr. Ina Pöhner from UEF's School of Pharmacy, successfully screened 1.56 billion drug-like molecules in less than 10 days – a process that traditionally would have required over six months using conventional methods. This breakthrough was achieved using HASTEN, an artificial intelligence-powered tool developed by Orion Pharma.
Technical Innovation and Performance
The screening process targeted two specific pharmacological targets: a bacterial chaperone protein and a viral kinase. HASTEN demonstrated remarkable efficiency by identifying 90% of the most promising drug candidates while only screening 1% of the entire compound library.
Dr. Tuomo Kalliokoski, who led HASTEN's development at Orion Pharma, explained the tool's sophisticated approach: "When presented with enough examples drawn from conventional docking, the machine-learning model can predict docking scores for other compounds in the library much faster than the brute-force docking approach."
Supercomputing Infrastructure
The research utilized two supercomputers, Mahti and Puhti, hosted by CSC - IT Center for Science. This powerful computing infrastructure was essential for processing the massive dataset and enabling the rapid screening capabilities demonstrated in the study.
Advancing Open Science
In a significant move to promote scientific advancement, the research team has made their entire dataset publicly available. This includes:
- The complete screening library
- Docking results for 1.56 billion compounds
- Benchmarking data for both pharmacological targets
This unprecedented data sharing will enable other researchers to develop and validate new computational drug discovery tools, potentially leading to further innovations in the field.
Impact on Drug Discovery
The development of HASTEN represents a significant leap forward in drug discovery efficiency. By dramatically reducing the time required for virtual screening, this technology could accelerate the identification of promising drug candidates and ultimately speed up the drug development process.
The research has been documented in the Journal of Chemical Information and Modeling, marking the first comprehensive comparison between ML-boosted docking screening and traditional brute-force approaches.
