Software
QSPRpred
QSPRpred is an open-source Python library designed to build, train, and deploy Quantitative Structure–Property/Activity Relationship (QSPR/QSAR) models. Developed in collaboration with the Computational Drug Discovery group at Leiden University, it streamlines machine learning workflows by automating dataset preparation, molecule descriptor calculation, and diverse model serialization.
GenUI
GenUI is a web-based platform and REST API framework designed to manage and visualize generative chemistry workflows, QSAR modeling, and chemical space exploration. It mainly attempts to bridge the gap between users and AI-driven molecular generation algorithms with an intuitive front-end user interface.
DrugEx
DrugEx (Drug Explorer) is a deep learning package for multi-objective *de novo* drug design that utilizes deep generative architectures (such as recurrent neural networks and transformers) alongside reinforcement learning. It enables researchers to generate novel molecule structures tailored to specific bioactivity thresholds, physical properties, and desirable target criteria.
Molpher & molpher-lib
Chemical space is a set of all possible chemically stable organic compounds. Its size is enormous, e.g. a number of structures with the molecular weight lower than 500 Da is estimated to be 1060. However, the majority of these compounds was not suggested, not to say synthesized and chacterized. Therefore, we have developed the method of molecular morphing [Hoksza D. et al. Journal of Cheminformatics 6 : 7 2014. DOI: 10.1186/1758-2946-6-7] that systematically generates chemical subspaces between two selected structures. This method is implemented in a freely available program Molpher and programatically accessible using the Python library molpher-lib.