CzeekD: Fragment-based de novo Drug Design System for Drug Discovery

presentation · 4 years ago
by Edmund Taylo (Kyoto Constella Technologies)
MarvinSketch JChem Base

It has been estimated that there is an enormous number of compounds that consist the vast chemical space and as such, the use of an optimal method to search for novel active compounds is generally required. In addition, the use of ligand-based virtual screening (LBVS) and structure-based virtual screening (SBVS) has their own limitations. Although rapid in its implementation, finding active compounds with novel structures is generally difficult when using LBVS. On the other hand, SBVS requires the 3-dimensional protein structure and a longer time for completion of a simulation task. Here we introduce a fragment-based drug design system, referred to as CzeekD, that enables de novo design of novel active compounds. It utilizes a unique technology referred as Chemical Genomics-based Virtual Screening (CGBVS) that scores the possibility of interaction between a compound and a target protein without requiring the protein 3-dimensional structure. The system identifies novel scaffolds based on a synthesis frame where building blocks from fragments libraries are combined. Furthermore, the generation of the fragment libraries and the synthesis of virtual compounds are based on the RECAP rules, hence only chemical structures with high chemical synthesis probability are created. We employed a newly developed Particle Swarm Optimization (PSO) algorithm to swiftly explore druggable molecular seeds and to efficiently find out diverse structures among the vast chemical space. The system also integrates two ChemAxon tools: JChem Base, that is required for rapid structure search and Marvin Sketch, that is used to draw custom structures . We have done implementations ourselves and obtained good results after confirmation with actual assays leading us to the application of substance patents.

Download slides here