Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.22893015We present QUPEX (QUantum Property Explorer), an interactive framework built on the open-source Streamlit platform for exploring the property spanned by QM7-X dataset and evaluating machine learning (ML) predictions. The currently indexed QM7-X working set comprises 6,950 unique molecular configurations (107,764 atoms across six elements: H, C, N, O, S, Cl) and 41,537 associated equilibrium geometries, drawn from the full $\approx$4.2 million-structure QM7-X corpus and made responsive for interactive use through a cached HDF5-to-Parquet indexing pipeline. The framework integrates four complementary pages that together form one continuous workflow for molecular data exploration, visualization, machine learning, prediction, and interpretation: dataset and geometry statistics, two- and three-dimensional structural and property visualization, unsupervised chemical-space exploration (PCA/t-SNE with Gaussian-Mixture clustering), and a machine-learning module for dipole-moment prediction. For the latter, a pretrained XGBoost regression pipeline over 40 tight-binding descriptors reproduces, on a held-out test set of 16,537 conformers, R\textsuperscript{2} = 0.9895, MAE = 0.0155, and RMSE = 0.0244 e$\cdot$\r{A} -- a fixed result we independently re-verified against the application's own code and data while preparing this manuscript. We describe the data-indexing strategy, the architecture and scientific role of each interactive module, and representative, disclosed-configuration use cases, carefully distinguishing implemented functionality, exploratory analyses, and fixed reproducible results throughout. The application and its source is available on GitHub. Code: Streamlit app: Preprint: FILES: -qm7x-eq.hdf5: -QMdesc_qm7x-eq_train.h5: -QMdesc_qm7x-eq_train.h5: -qued-model_dip.pkl:
Keyword(s):
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)