The Impact of Supervised Learning Methods in Ultralarge High-Throughput Docking.

Area of Science:

• Computational chemistry
• Drug discovery
• Machine learning applications

Background:

• Structure-based virtual screening is crucial in computational drug discovery.
• Large-scale virtual screening campaigns identify novel drug hits but are often inaccessible.
• Machine learning (ML) offers a solution to accelerate hit identification in vast chemical spaces.

Purpose of the Study:

• To illustrate the challenges of screening large chemical databases.
• To provide an overview of ML-accelerated protocols for virtual screening.
• To focus on supervised learning methods in this context.

Main Methods:

• Review of machine learning-based protocols for virtual screening.
• Discussion of supervised learning techniques applied to large-scale screening.
• Analysis of computational time reduction strategies.

Main Results:

• ML-based protocols significantly reduce computational time for virtual screening.
• These methods enable efficient exploration of ultralarge chemical spaces.
• Novel hits can be identified more rapidly through ML acceleration.

Conclusions:

• Machine learning is revolutionizing virtual screening by enhancing speed and accessibility.
• Supervised learning methods show particular promise for accelerating drug discovery.
• Future studies should explore further insights and applications of ML in this field.