Virtual screening identifies hits from extensive chemical space

Virtual screening (VS) is a computational technique used in drug discovery to search real or virtual libraries of small molecules in order to identify potential hit candidates.

  • Search against large chemical space (from millions to billions)
  • Efficient screening with high success rate
  • Integrated services to provide rational starting point for Hit Identification, Lead Optimization and Medicinal Chemistry
  • Reduce real laboratory experiments and accelerating the drug discovery process in a more efficient and economical way
  • 3.4Bn compound ‘make-on-demand’ library provides rapid access to real samples, shipped in 4-6 weeks with successful rate of up to 80%

Virtual Library

WuXi AppTec offers access to virtual chemical spaces built from our library of novel drug-like scaffolds. Due to carefully selected building block supply and proven chemistry, these libraries consist of molecules that can be quickly synthesized on demand.

Core-Based Library

  • Create reaction scheme for interested molecular scaffold (core)
  • Identify/search available building block sets for reaction scheme
  • Enumerate compounds for reaction scheme of each scaffold
  • Create diverse compound set to form core-based focused library

Synthetically Feasible Virtual Library

The library covers rigorously validated chemical space of over 200 million virtual compounds based on optimized one-step/one-pot reactions. It is a combination of WuXi AppTec’s chemistry expertise and cutting edge software technologies, resulting in a chemical space over 200 million virtual molecules that can be mined and synthesized within a few weeks.

GalaXi (WuXi AppTec & BioSolveIT) – Virtual Space Design

Billions of accessible small molecules search with vHTS or with BioSolveIT Infinisee


Machine learning empowered virtual high-throughput screening

Advances in ML-empowered vHTS is allows us to access a rapidly expanding and diverse chemical space, to discover potent novel molecules with accelerated timeline and reduced cost.

Key features of ML-empowered vHTS platform:

  • Vastly expand the chemical space that is limited by real compound collections
    • Ability to process large chemical spaces (from millions to billions)
    • In silico screening with on-the-fly enumeration and make-on-demand virtual libraries
  • Reduce real laboratory experiments and transform drug discovery in a more efficient and economical way
    • Provide rational starting points for integrated services
    • Evaluate target binding with high success rate of activity
    • vHTS trained ML models can create generative designs to expedite hit-to-lead and lead optimization

Using output from hit-finding campaigns or other existing data, ML-empowered vHTS has shown to identify diverse, drug-like and easily-synthesizable molecules that are different from the original screening libraries or datasets

Predict phenotypic activities to discover novel antibiotics 

  • Deep neural network to predict antibiotic activity in molecules that are structurally different from existing antibiotics
  • Compound demonstrated efficacy in mice infection model2

1] J Med Chem. 2020. doi: 10.1021/acs.jmedchem.0c00452 [2] Cell. 2020 Feb 20;180(4):688-702.e13. doi: 10.1016/j.cell.2020.01.021.