Table 2 Comparison of PDNET, DeepDist, and SDP methods in terms of mean lDDT values over all residue pairs in each dataset.
From: Enhancing protein inter-residue real distance prediction by scrutinising deep learning models
Test | Prediction | Mean | MAE for \(D_{ij}<16\) | MAE for \(D_{ij}<36\) | ||||
|---|---|---|---|---|---|---|---|---|
Dataset | Method | lDDT | \(S_{ij}\ge 1\) | \(S_{ij}\ge 12\) | \(S_{ij}\ge 24\) | \(S_{ij}\ge 1\) | \(S_{ij}\ge 12\) | \(S_{ij}\ge 24\) |
CASP13.31 | PDNET | 0.326 | 3.89 | 5.37 | 5.77 | 4.55 | 5.12 | 5.41 |
SDP-X | 0.352 | 3.33 | 4.55 | 4.86 | \(*\)3.37 | 4.46 | 4.67 | |
SDP-L | 0.393 | 3.02 | 4.09 | 4.34 | 3.88 | \(*\)4.37 | \(*\)4.58 | |
DeepDist | 0.503 | 1.73 | 1.94 | 1.97 | 6.39 | 7.12 | 7.33 | |
SDP-Y | 0.475 | 1.74 | 2.17 | 2.20 | 6.36 | 7.16 | 7.38 | |
SDP-Z | 0.540 | 1.66 | 1.92 | 2.01 | 8.40 | 9.48 | 9.58 | |
SDP-S | \(*\)0.569 | \(*\)1.49 | \(*\)1.82 | \(*\)1.85 | 7.88 | 8.91 | 9.25 | |
CAMEO.131 | PDNET | 0.361 | 4.00 | 5.94 | 6.59 | 4.75 | 5.52 | 5.93 |
SDP-X | 0.396 | 3.14 | 4.52 | 5.04 | 3.79 | 4.29 | 4.55 | |
SDP-L | 0.451 | 2.67 | 3.8 | 4.11 | \(*\)3.64 | \(*\)4.23 | \(*\)4.37 | |
DeepDist | 0.527 | 1.53 | 1.88 | 1.92 | 6.92 | 7.93 | 8.21 | |
SDP-Y | 0.516 | 1.53 | 1.93 | 1.97 | 6.68 | 7.75 | 8.02 | |
SDP-Z | 0.559 | 1.56 | 1.92 | 1.93 | 9.32 | 10.90 | 10.62 | |
SDP-S | \(*\)0.596 | \(*\)1.36 | \(*\)1.67 | \(*\)1.73 | 8.26 | 9.6 | 9.96 | |
CAMEO.144 | PDNET | 0.420 | 3.27 | 4.71 | 5.15 | 3.99 | 4.56 | 4.96 |
SDP-X | 0.450 | 2.78 | 3.92 | 4.26 | 3.47 | 3.98 | 4.25 | |
SDP-L | 0.506 | 2.25 | 3.06 | 3.23 | \(*\)3.31 | \(*\)3.81 | \(*\)4.15 | |
DeepDist | 0.570 | 1.48 | 1.63 | 1.66 | 6.41 | 7.50 | 8.32 | |
SDP-Y | 0.567 | 1.37 | 1.71 | 1.75 | 6.33 | 7.49 | 8.21 | |
SDP-Z | 0.595 | 1.49 | 1.71 | 1.68 | 9.10 | 10.04 | 10.85 | |
SDP-S | \(*\)0.631 | \(*\)1.24 | \(*\)1.54 | \(*\)1.55 | 7.79 | 9.24 | 10.1 | |
