“…Moreover, a correlation study was conducted on eleven further experimental systems. These included a small well-described system (PDB ID: 1CRN [55]; Figure S5), a large system of Salinosporamide A complexed with yeast 20S proteasome (PDB ID: 2FAK [56]; Figure S6), homodimeric hemoglobin (PDB ID: 3QOB [57], Figure S7), a mediumsized system E. Coli DNA gyrase subunit B (PDB ID: 4DUH [58]; Figure S8), IFN alpha8 (PDB ID: 6JHD [59]; Figure S9), crystal structure of HL homo-diabody (PDB ID: 6KR0 [60]; Figure S10), a cryo-EM structure of the human PA200 and PA200-20S complex (PDB ID: 6KWY [61]; Figure S11), S109 in complex with CRM1-Ran-RanBP1 (PDB ID: 6LQ9 [62]; Figure S12), and a structure of NHP D11A.F2 Fab (PDB ID: 6XLZ [63]; Figure S13), as well as both Ala and Val MnSOD models (Figures S14 and S15), based on the X-ray diffraction by Azadmanesh, et al [64] and studied by Broz et al [65]. All model predictions exhibit MAE values similar to the test dataset predictions, ranging from 12.24 • to 28.26 • for φ errors and from 16.70 • to 77.66 • for ψ errors (Table S3), and the I model could generally classify the correct secondary structure.…”