The Fragment Network: A Chemistry Recommendation Engine Built Using a Graph Database.

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Discovery

Background:

• Fragment-based drug discovery (FBDD) requires exploring chemical space around initial fragment hits.
• Identifying similar compounds from large corporate or commercial collections is crucial for hit validation.
• Existing methods for chemical space exploration can be inefficient or lack intuitive organization.

Purpose of the Study:

• To introduce the Fragment Network, a novel graph database for efficient chemical space searching.
• To describe the algorithms for constructing and querying the Fragment Network.
• To demonstrate the utility of the Fragment Network in drug discovery contexts.

Main Methods:

• Development of a graph database (Fragment Network) to represent chemical compounds and their relationships.
• Implementation of algorithms for constructing the network from chemical databases.
• Design of search algorithms to query the network based on a compound of interest.
• Utilizing medicinal chemistry databases to inform the sorting and grouping of search results.

Main Results:

• The Fragment Network enables efficient searching of chemical space around fragment hits.
• Search results are chemically intuitive, grouped by substitution patterns.
• Results are meaningfully sorted based on transformation observations in medicinal chemistry databases.
• Demonstrated practical applications in drug discovery scenarios.

Conclusions:

• The Fragment Network provides an efficient and intuitive tool for hit validation in FBDD.
• The graph database approach facilitates the exploration of SAR around fragment hits.
• This method aids medicinal chemists in identifying promising lead compounds.