Skip to main content

Table 1 Discriminating power of individual features

From: Accurate discrimination of conserved coding and non-coding regions through multiple indicators of evolutionary dynamics

Long CST training set Short CST training set
Feature AUC Feature AUC
CPS-ratio 0.9128 SpectAlign 0.8466
SpectrAlign 0.9113 CPS-best 0.8322
Ns/Nns-best 0.8538 CPS-ratio 0.8313
CPS-best 0.8532 Aasim-best 0.8259
GC-target 0.8263 Ns/Nns-best 0.8178
GC-probe 0.8231 GC-probe 0.7796
Ns/Nns-ratio 0.7948 GC-target 0.7727
AAsim-best 0.7893 AAID-best 0.7641
Stop-best 0.7854 Ns/Nns-ratio 0.7370
Codon-sim-ratio 0.7674 Aasim-ratio 0.6406
AAID-best 0.7222 AAID-ratio 0.6184
AAsim-ratio 0.6881 Codon-sim-ratio 0.6160
AAID-ratio 0.6626 GFB-ntID 0.6159
GFB-length 0.5587 GFB-length 0.6106
Tv/subs 0.5324 Stop-delta 0.6030
Stop-delta 0.5183 Stop-best 0.5915
GFB-ntID 0.5166 Tv/subs 0.5119
  1. For each feature, Area Under Curve (AUC) calculated from ROC curves calculated with training sets was used as a measure of discriminating power of the feature (see results)