GlyContact analyzes glycan 3D structures at scale.

Glycans are branched, structurally diverse, and highly flexible biomolecules. These characteristics make glycoanalytics and structural characterization challenging, resulting in often unclear structure-to-function relationships. GlycoShape, currently the largest open-access database of glycan 3D structures from molecular dynamics (MD) simulations, provides an opportunity to fill this information gap. Here, we present GlyContact, an open-source Python package designed and developed to retrieve, process, and analyze glycan 3D structures, from MD, NMR, or X-ray crystallography. We demonstrate that GlyContact can (i) unveil the impact of sequence context on glycan motif structure, (ii) yield a predictive understanding of motif flexibility and surface accessibility on lectin-glycan binding, which improved lectin-binding prediction by ~ 7%, and (iii) accurately predict torsion angle distribution between disaccharides using von Mises graph neural networks. We envision that GlyContact will allow researchers to explore glycan structures within their 3D space, obtaining insights into their biological functions. GlyContact is available open-access at https://github.com/lthomes/glycontact .