FIG. 2.

Automated identification of protein conformational states across the PDB archive. (a) All chains of a given UniProt accession (100% sequence identity) are assigned to segments based on their overlap with the reference UniProt sequence. Non-overlapping sequences are grouped into separate segments. (b) Chains are superposed to all other chains within their assigned segment. (c) Chain–chain GLOCON scores are calculated for all polypeptides within a segment (refer to Ref. 89 for formal definition) before (d) agglomerative clustering is performed. The results are displayed in 3D on PDBe-KB aggregated views of proteins pages.