PDB-wide identification of biological assemblies from conserved quaternary structure geometry
Résumé
Protein structures are key to understanding bio-molecular mechanisms and diseases, yet their interpretation is hampered by limited knowledge of their biologically relevant quaternary structures (QSs). A critical challenge in obtaining QSs from crystallographic data is to distinguish biological interfaces from crystal packing contacts. We tackled this challenge with two strategies for aligning and comparing QS states, both across homologs (QSalign), and across data repositories (QSbio). QS conservation across homologs was a remarkably strong predictor of biological relevance and allowed annotating of >80,000 biological QS states. QS conservation across methods enabled us to create a meta-predictor, QSbio, from which we inferred confidence estimates for >110,000 assemblies in the Protein Data Bank, which approach the accuracy of manual curation. Based on the dataset obtained, we analyzed interaction interfaces among pairs of structurally conserved QSs. This revealed a striking plasticity of interfaces, which can maintain a similar interaction geometry through widely different chemical properties.
Domaines
Bio-informatique [q-bio.QM]Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...