Congestion game scheduling for virtual drug screening optimization.

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Chemical diversity of hits is crucial in virtual drug screening, alongside predicted activity.
• Interim results are valuable for guiding research and should also exhibit diversity.

Purpose of the Study:

• To develop a method for obtaining a diverse set of virtual screening hits rapidly.
• To optimize task scheduling in high-performance computing for virtual drug screening.

Main Methods:

• A mathematical model of task scheduling as a congestion game was developed for computational nodes.
• The model addresses heterogeneous environments with uncertain workloads and processing times.
• Equilibrium solutions were sought to balance interim hit numbers and chemical diversity.

Main Results:

• Computational experiments using the GDB-9 organic molecules database demonstrated feasibility.
• The game-based scheduling algorithm outperformed two common heuristics in hit discovery rate and chemical diversity at early stages.
• Equilibrium solutions achieved high values for the social utility metric.

Conclusions:

• The proposed game-based scheduling approach effectively enhances both the speed and chemical diversity of virtual drug screening hits.
• This method offers a practical solution for optimizing resource allocation in high-performance computational systems for drug discovery.