alvaBuilder

alvaDesc is a cheminformatics software for the calculation and analysis of molecular descriptors, fingerprints, and structural patterns for QSAR, QSPR, read-across, and machine learning applications. It computes more than 5,000 molecular descriptors (0D–3D), including constitutional, topological, geometrical, electronic, physicochemical, and fragment-based descriptors. The software also generates molecular fingerprints and structural pattern counts for similarity analysis, clustering, and classification. Integrated tools support descriptor filtering and correlation analysis for robust and reproducible modeling. alvaDesc integrates seamlessly with KNIME and Python, enabling efficient connection to external data analysis and machine learning workflows. It is widely used in academic and industrial research and supported by extensive documentation and scientific publications.

alvaModel is a software tool for building, validating, comparing, and applying QSAR and QSPR models. It supports regression and classification workflows based on molecular descriptors and fingerprints, with a strong focus on model transparency, interpretability, and scientific robustness. The software includes multiple data splitting strategies, variable selection methods, modeling algorithms, and comprehensive internal and external validation procedures. alvaModel provides diagnostic plots, applicability domain analysis, and model comparison tools to support the identification of reliable and predictive models. Designed according to best practices in chemometrics, alvaModel facilitates the development of interpretable models consistent with the OECD principles for QSAR validation, making it suitable for research and regulatory-oriented applications. The graphical interface guides users through the entire modeling workflow while allowing full control over each modeling step.