Skip to main content

Table 1 Training vs. testing groups: native recognition test

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

  Total sets # of
Natives
# RMSD
< 2Å
# RMSD
< 5Å
Ranknat Cα RMSDbest Znat CCnat FEnat (%)
Training group 154 57 112 136 47.1 2.10 1.587 0.478 32.9
Testing group 77 38 60 70 13.4 1.62 1.805 0.519 36.6
  1. Comparison of LoCo performance for native structure recognition on both the training and testing groups of decoy sets is shown. Results are roughly comparable, though the testing group did somewhat better across the board. Natives, RMSD < 2Å and RMSD < 5Å refer to the number of times the best-scoring structure in a particular group is either the native structure (0Å Cα RMSD) or within 2Å or 5Å Cα RMSD from the native structure, respectively. All other measures are averages over every decoy set in the group. Definitions of measures used are provided in the Performance Measures subsection at the end of Methods. In summary, lower scores are better for Ranknat and RMSDbest; higher scores are better for all the other measures..