Table 1.
Benchmarking accessibility predictors
Mem Proteins | W-S Proteins | ||||||
---|---|---|---|---|---|---|---|
All | Z:0-10 | Z:10-22 | Z: > 22 | Membr | non-Membr. | All | |
MPRAP | 0.45 | 0.47 | 0.40 | 0.49 | 0.45 | 0.46 | 0.55 |
ACCPRO | 0.41 | 0.19 | 0.34 | 0.53 | 0.20 | 0.47 | 0.63 |
SABLE | 0.29 | 0.08 | 0.24 | 0.46 | 0.11 | 0.38 | 0.55 |
TMX | 0.21 | 0.35 | 0.19 | 0.13 | 0.32 | 0.14 | - |
ACCPRO+TMX | 0.39 | 0.33 | 0.30 | 0.53 | 0.31 | 0.47 | - |
ASAPmem | 0.12 | 0.05 | 0.14 | 0.11 | 0.08 | 0.13 | - |
ASAPextmem | 0.40 | 0.26 | 0.41 | 0.47 | 0.27 | 0.44 | - |
ASAPglob | 0.32 | 0.11 | 0.25 | 0.46 | 0.11 | 0.36 | - |
A comparison of the performance for different accessibility predictors using two-state predictions in water-soluble (W-S) proteins and membrane (Mem) proteins. The reported values are the Matthew correlation coefficients for identifying buried residues in a binary alphabet. Analysis was performed the entire protein (whole) or regions in the membrane either divided by Z-coordinate or by membrane definitions from OPM. Due to computational limitations only three predictors were applied on the water-soluble dataset.