| Â | RB1 | RB10 | RMSDdecoy | Zdecoy | CCdecoy | FEdecoy (%) | log(PB1) | log(PB10) |
|---|
LoCo
| 154.8 | 5.6 | 3.51 | 0.938 | 0.505 | 31.4 | -0.864 | -1.640 |
RAPDF
| 152.2 | 27.7 | 4.02 | 0.878 | 0.479 | 30.8 | -0.818 | -1.604 |
dDFIRE
| 136.5 | 18.9 | 3.75 | 1.014 | 0.536 | 33.3 | -0.896 | -1.592 |
DOPE
| 97.4 | 52.0 | 4.21 | 0.764 | 0.466 | 25.9 | -0.717 | -1.409 |
RF_HA_SRS
| 122.5 | 56.9 | 3.45 | 0.896 | 0.493 | 33.1 | -0.881 | -1.526 |
- Decoy discrimination performance for LoCo is compared to that of four widely-used all-atom potentials. All reported measures are averages over the 77 decoy sets in the final testing group. LoCo outperforms all others at RB10 and log(PB10). LoCo is beaten by all four at RB1. LoCo places 2nd among all potentials at RMSDdecoy, Zdecoy and CCdecoy. It places 3rd in FEdecoy and log(PB1). When the sum of all rankings among these five potentials is considered, LoCo places 2nd overall. All metrics are defined in Performance measures at the end of Methods.
