Anti-Bacterial Library / Anti-Viral Library

Anti-Bacterial Library and Anti-Viral Library in One Set

Medicinal and Computational Chemistry Dept., ChemDiv, Inc., 6605 Nancy Ridge Drive, San Diego, CA 92121 USA, Service: +1 877 ChemDiv, Tel: +1 858-794-4860, Fax: +1 858-794-4931, Email: [email protected]


Molecules in this set are based on the literature analysis of both targets and their respective ligands, specifically 5,800 known molecules with reported anti-bacterial and anti-viral activities. In order to assemble the set, we used a) Kohonen self-organizing map, b) pharmacophore analysis, c) bioisosteric replacement and d) internal peptidomimetics in order to select initial library of ca. 50,000 drug-like molecule. These were further prioritized based on relevance to the target/ligand of interest, IP potential, Lipinski-rule-of-five parameters and synthetic feasibility to yield final set of 10,000 molecules. Approximately 10% of these compounds were proven to possess relevant biological activity in our internal screening and via collaborative efforts.

Example: Application of Kohonen self-organizing maps to selection of compounds with potential anti-bacterial activity.
Structures in the training set: 5806 antibacterial agents from the Prous Integrity database
Structures in the test set: 10100 compounds from the focused library
Approach: analysis of multidimensional property space of antibacterial compounds and focused library using Kohonen self-organizing maps (for description of the method, see [K. V. Balakin, Y. A. Ivanenkov, N. P. Savchuk, A. A. Ivashchenko, S. Ekins. Comprehensive Computational Assessment of ADME Properties Using Mapping Techniques // Curr. Drug Disc. Techn. 2005. V. 2. 99-113.])
Computational tool: SmartMining program, ver. 1.01
Description of Experiment:

  1. A total of 90 molecular descriptors were calculated for both training and test sets.
  2. 15 descriptors (table 1) were selected using Principal Component Analysis (PCA) and used for generation of Kohonen map.
  3. A Kohonen map of the training dataset (5806 antibacterial agents) was generated using 15 descriptors (Fig. 1); this map can be used for assessment of “antibacterial-likeness” of chemical compounds.
  4. 4. Compounds from the focused library (test set of 10113 compounds) were placed on the same map (Fig. 2).

Compounds from the focused library (green area, Fig. 2) are located in regions of the map which contain antibacterial agents (blue area, Fig. 1). Therefore, the molecular properties of compounds from the focused library are similar to properties of the antibacterial agents.

Download PDF Presentation

Request Library

Share Library