Projects

The projects below are grouped into three umbrella areas that reflect the main scientific directions of my work. Each area combines method development, software implementation, and application-oriented research.

Reaction Informatics

Reaction Informatics

Methods and software for reaction representation, reaction curation, rule extraction, classification, benchmarking, and mechanism-aware modeling.

Core framework

SynKit provides graph-based reaction modeling and analysis, while SynRBL supports reaction rebalancing and dataset curation.

Data and learning layer

SynTemp, SynCat, SynRXN, and SynEdu cover rule extraction, reaction classification, benchmarking, and educational resources.

Drug Discovery

Drug Discovery

Reproducible computational workflows for molecular property prediction, docking, model evaluation, and practical structure-based discovery.

Predictive modeling

ProQSAR focuses on small-data QSAR workflows, rigorous validation, and fit-and-use predictive models.

Structure-guided workflows

ProDock automates docking analysis, and KGG adds knowledge-guided graph learning for molecular property prediction.

Biomolecular Modeling

Biomolecular Modeling

Computational methods for peptide chemistry, peptide--protein interaction analysis, confidence-aware scoring, and biomolecular design workflows.

Modeling and analysis

PepKit supports peptide chemistry, graph-based modeling, structural querying, and post-processing workflows.

Scoring and design support

PepQ provides confidence-aware scoring and ranking of peptide--protein complexes for computational design.