Theoretical and Computational Chemistry

Lib-INVENT: Reaction Based Generative Scaffold Decoration for in silico Library Design

Abstract

Due to the strong relationship between desired molecular activity to its structural core, screening of focused, core sharing chemical libraries is a key step in lead optimisation. Despite the plethora of current research focused on in silico methods for molecule generation, to our knowledge, no tool capable of designing such libraries has been proposed. In this work, we present a novel tool for de novo drug design called Lib-INVENT. This is capable of rapidly proposing chemical libraries of compounds sharing the same core while maximising a range of desirable properties. To further help the process of designing focused libraries, the user can list specific chemical reactions that can be used for the library creation. Lib-INVENT is therefore a flexible tool for generating virtual chemical libraries for lead optimisation in a broad range of scenarios. Additionally, the shared core ensures that the compounds in the library are similar, possessing desirable properties and can be also synthesized under the same or similar conditions.

Version notes

Updated figures in both the main paper and the supporting information. Included further statistics about the training and generated compounds and corrected minor numbering errors.

Content

Thumbnail image of Lib-INVENT.pdf

Supplementary material

Thumbnail image of Lib-INVENT Supporting information.pdf
Supporting information
Supplementary methods, figures and tables for the publication.

Supplementary weblinks

Preprocessing code
GitHub repository holding the code necessary for preprocessing of the datasets including reaction based slicing.
Main code repository
GitHub repository holding the model definitions, pretrained prior models and code necessary for model training. The data used in the publication is also included.