Skip to main content

Table 8 LoCo variation: decoy discrimination

From: LoCo: a novel main chain scoring function for protein structure prediction based on local coordinates

  RB1 RB10 RMSDdecoy Zdecoy CCdecoy FEdecoy (%) log(PB1) log(PB10)
LoCo BEST 25.5 5.6 3.01 1.005 0.517 33.0 -0.982 -1.654
LoCo WORST 175.0 53.0 4.09 0.748 0.374 24.0 -0.665 -1.432
LoCo AVERAGE 112.5 26.5 3.68 0.909 0.481 29.9 -0.778 -1.562
LoCo CHOSEN 154.8 5.6 3.51 0.938 0.505 31.4 -0.864 -1.640
DFMAC 108.9 13.8 3.64 1.024 0.533 31.6 -0.825 -1.586
RF_CB_SRS_OD 172.8 52.5 4.11 0.914 0.457 28.4 -0.761 -1.524
ProSa 2003 118.2 24.8 3.82 0.931 0.493 32.3 -0.755 -1.650
Server BEST 90.7 25.3 3.81 0.725 0.416 27.3 -0.671 -1.434
Server WORST 231.3 71.3 5.83 -0.001 0.165 7.6 -0.359 -1.043
Server AVERAGE 159.3 41.5 4.64 0.494 0.328 20.2 -0.544 -1.306
  1. Best, worst and average performance for LoCo across all 84 parameter sets tested is compared with the chosen LoCo parameter set, the three best-performing of the other potentials, and the best, worst and average performance of all 26 remaining potentials from the Jernigan Lab server. All best, worst, and average values are for each individual performance measure; no single set contained all those values. All reported measures are averages over the 77 decoy sets in the final testing group. Lower scores are better for RB1, RB10, RMSDdecoy, log(PB1) and log(PB10). Higher scores are better for Zdecoy, CCdecoy and FEdecoy. The average performance for LoCo among all 84 parameter sets exceeds all other functions except DFMAC in RMSDdecoy and log(PB1). The LoCo average betters all other functions except DFMAC and ProSa 2003 in log(PB10). All metrics are defined in Performance measures at the end of Methods.