Skip to main content
. 2024 Jan 4;25(1):bbad496. doi: 10.1093/bib/bbad496

Table 3.

Coverage of structural protein classes according to PDB and AF

Structural class Total protein count Proteins [have PDB] (average PDB coverage, %) Protein count [all predictions w/ AF] (average pLDDT score) Protein count [high acc. predictions w/ AF] (average pLDDT score)
Mainly alpha 6406 2830 (58.5) 6294 (76.5) 1735 (89.3)
Mainly beta 4711 2395 (56.1) 4577 (77.9) 1220 (89.7)
Mixed alpha–beta 6696 3268 (65.3) 6585 (78.7) 2367 (91.0)

Proteins in reference proteome are classified according to the data provided by CATH-Gene3D. Proteins [have PDB] (average PDB coverage, %): The number of UniProt entries that belong to each class and have at least a PDB structure and average PDB coverage of the class calculated by taking the averages of UniProt entry’s corresponding PDB coverages (see Methods for coverage calculations). Protein count [all predictions w/ AF] (average pLDDT score, 0–100): The number of UniProt entries that belong to each class and have a predicted 3D structure produced by AF. In an AF prediction file, pLDDT scores of each residue were averaged and referred to as prediction accuracy of this corresponding UniProt entry. Protein count [high acc. predictions w/ AF] (average pLDDT score, 0–100): the number of UniProt entries that belong to each class and have a predicted 3D structure that 85% of the residues predicted with ≥70% pLDDT score. Only the UniProt entries within a class predicted with HA are considered and calculated by taking the averages of the UniProt entry’s corresponding AF prediction accuracies.