DPCstruct Metaclusters

⬡ Domain Classification in Structural Space

The DPCstruct pipeline applied the Density Peak Clustering (DPC) algorithm to roughly 15 million representative AlphaFold2- predicted structures from the Foldseek Cluster database (a representative set of the AlphaFold Database, AlphaFoldDB v4.0), producing 28,246 structural metaclusters without any manual annotation. Because protein structure is more conserved than sequence over evolutionary time, DPCstruct recovers remote homologs, including proteins sharing a common fold at less than 30% sequence identity, that purely sequence-based methods typically miss.

Use the toggle buttons to switch between the Properties view (intrinsic structural metrics and Pfam annotations), the CATH fold annotations, or the SCOP fold annotations. Click any MCID to open its detail page, where you can explore the sequences and launch the integrated PDBe-Mol* 3D viewer.

CATH Fold MCID DPC Target Q.Range T.Range Q.Cov T.Cov Q.TM-Score T.TM-Score A.TM-Score lDDT pIdent
1.10.1450 MC10171 A0A452TJB2_10171 1-86 68-153 0.99 0.46 0.96 0.45 0.45 0.92 94.1
3.40.532 MC1013 A0A3Q7Q5K1_1013 3-207 13-118 0.99 0.88 0.39 0.64 0.64 0.44 10.6
1.10.4100 MC10068 A0A6L7NGA6_10068 3-266 24-283 0.84 0.85 0.76 0.78 0.78 0.66 23.0
3.30.260 MC10004 A0A484B1D0_10004 1-91 127-401 1.0 0.54 0.99 0.18 0.18 0.98 27.6
Legend
  • Agreement between CATH Folds and DPCstruct Metacluster Domains:
  • CATH Fold: CATH fold identifier.
  • MCID: DPCstruct metacluster identifier.
  • DPC Target: Target DPCstruct protein domain.
  • Q./T.Range: Amino acid coordinate ranges for query and target.
  • Q./T.Cov: Coverage metrics indicating how much of each domain was covered in the alignment.
  • A.TM-Score: Template Modeling score of the structural alignment (0-1).
  • lDDT: Local Distance Difference Test score.
  • pIdent: Percent sequence identity between query and target.