QSPRpred

QSPRpred provides a modular Python API for quantitative structure-property relationship (QSPR) modeling of bioactivity data sets, with direct serialization of data sets and models to ensure reproducibility and enable prediction on new compounds from SMILES strings.

Key Features:

• Modular API: Enables intuitive description of different parts of a modeling workflow using pre-implemented components and integration of customized implementations in a plug-and-play manner.
• Direct Serialization: Data sets and models are directly serializable, saving all required data pre-processing steps to make predictions on new compounds directly from SMILES strings.
• Multi-task Modeling: Supports multi-task modeling for simultaneous prediction of multiple endpoints.
• Proteochemometric Modeling: Includes support for proteochemometric modeling that incorporates both compound and target information.
• Integration with Popular Packages: Integrates with many popular packages that extend beyond QSPR modeling.

Scientific Applications:

• Bioactivity Data Analysis: Facilitates analysis of bioactivity data sets for structure-property relationship studies.
• QSPR Model Creation: Supports creation of quantitative structure-property relationship models from curated experimental data.
• Model Deployment: Enables transfer of trained models into practice for prediction on new chemical compounds.
• Benchmarking: Allows comparison of diverse sets of models through its component-based architecture.

Methodology:

QSPRpred employs a modular Python API architecture with pre-implemented components for workflow description, automated serialization of data sets and models with embedded pre-processing steps, and support for multi-task and proteochemometric modeling approaches.