Perspectives on current approaches to virtual screening in drug discovery

Area of Science:

• Computational chemistry
• Drug discovery
• cheminformatics

Background:

• Virtual screening (VS) has been a key approach in drug discovery for 20 years.
• Billions of compounds are screened, with many successful VS examples reported.
• VS methods are continuously advancing, incorporating machine learning and physics-based techniques.

Purpose of the Study:

• To review recent VS applications in drug discovery.
• To discuss results from the Critical Assessment of Computational Hit-finding Experiments (CACHE) challenge.
• To analyze cost, open-source options, chemical space coverage, and library selection in VS.

Main Methods:

• Examination of recent virtual screening case studies.
• Analysis of prospective hit-finding results from the CACHE challenge.
• Evaluation of cost-effectiveness and open-source platforms for VS.

Main Results:

• Ultra-large libraries (ULL) of billions of molecules can be screened using advanced VS techniques.
• Prospective ULL VS campaigns have yielded potent and novel hits across various targets.
• Effective VS often involves using smaller, focused libraries tailored to specific targets.

Conclusions:

• Virtual screening is most effective when conducted in a fit-for-purpose manner, with careful selection of chemical space.
• While ULL screening is advancing, traditional focused libraries remain valuable.
• Further development of VS methods is needed to address more challenging drug targets.