Over 20 years of reaction access systems from MDL: a novel reaction substructure search algorithm

Area of Science:

• Chemical informatics
• Computational chemistry
• Database management

Background:

• MDL has evolved its reaction database management systems over time.
• Previous systems include REACCS and MDL ISIS/Host Reaction Gateway.
• The latest system is MDL Relational Chemistry Server, utilizing Oracle data cartridge technology.

Purpose of the Study:

• To detail the evolution of MDL's reaction substructure search (RSS) algorithms.
• To introduce a novel RSS algorithm.
• To compare the performance of new and existing RSS algorithms.

Main Methods:

• The article discusses the evolution of system architecture and RSS algorithms.
• A novel RSS algorithm based on depth-first search is described.
• This algorithm leverages reaction-specific information like reacting centers and atom-atom mapping (AAM).

Main Results:

• The new RSS algorithm is implemented in the MDL Relational Chemistry Server.
• It enables precise retrieval of reaction instances.
• It significantly minimizes irrelevant search results.

Conclusions:

• The novel depth-first search-based RSS algorithm enhances reaction database searching.
• The algorithm's use of reaction-specific data improves search accuracy and relevance.
• MDL's reaction database systems continue to advance in capability.