Table 1

Feature sets to discriminate OL from others

Only pairwise classifiers that ranked at higher than 100th in both lists are included. Triplet-wise classifiers are included only when they are ranked at higher than 50th in both lists. For any feature set, ες denotes the error of the optimal classifier for the feature set, and Δ(ες) denotes the largest decrease in error for the full feature set relative to all of its subsets. LOO is computed by designing n classifiers from sample subsets formed by leaving out one data point at a time, and then each classifier is applied to the left-out point, and the estimator LOO is 1/n times the number of errors by the n classifiers.

Gene namesεςΔ(ες)LOO
Transducin β2 subunit 20.02070.00
Transducin β10.08900.08
Growth factor receptor-bound protein 2 (GRB2)0.11150.12
Cyclin D3SMARCA40.07120.09860.04
Follitropin receptorThymosin β100.07600.09210.04
MUC18Transducin β1GRB20.01560.02130.00
MUC18Transducin β1RXR-β0.03100.02700.04
MUC18Transducin β1Clusterin0.03190.02600.00
MUC18Transducin β1α 1 catenin0.03640.02150.04
MAP kinase 10SMARCA4Neuronal acetylcholine0.03880.05000.00
Receptor α3
MRPGRB2Erythropoietin receptor0.03880.02740.04
MAP kinase 10Follitropin receptorNeuronal acetylcholine0.03970.05310.00
Receptor α3
MUC18GRB2Erythropoietin receptor0.04060.02560.04
ClusterinISGF3 γErythropoietin receptor0.04310.04620.00