Sheffield Chemoinformatics

Graph-based spectral geometry descriptors that combine molecular shape and electrostatic information in a compact, alignment-free representation. Includes tools for descriptor generation, similarity searching, and virtual screening benchmarks.

Visualisation tool for exploring lead optimisation series using reduced graph representations of molecules. Includes the code, datasets, and workflows described in the accompanying publication.

Code, datasets, and results accompanying the study "Effect of Missing Data on Multitask Prediction Methods". Provides implementations of Random Forest, Deep Neural Network, and Macau models for evaluating the impact of missing data on multitask QSAR performance.

Alignment-free 3D molecular shape descriptors based on spectral and diffusion geometry. Provides tools for computing Laplace-Beltrami spectra, Wave Kernel Signatures (WKS), and global shape descriptors for molecular similarity and virtual screening.