Predicting gene function using hierarchical multi-label decision tree ensembles

BMC Bioinformatics

Table 3 Comparison of between Clus-HMC-Ens and the MouseFunc systems

Subset	CLUS-HMC-ENS	BSVM⁺	KLR	CSVM	GENE FAS	GENE MANIA	KIM	Funckenstein
BP_3-10	0.695	0.808⊕	0.581⊖	0.588⊖	0.715⊕	0.873⊕	0.813⊕	0.790⊕
BP_11-30	0.748	0.808⊕	0.741⊖	0.659⊖	0.767⊕	0.849⊕	0.822⊕	0.796⊕
BP_31-100	0.831	0.874⊕	0.846⊕	0.778⊖	0.780⊖	0.872⊕	0.851⊕	0.880⊕
BP_101-300	0.823	0.853⊕	0.845⊕	0.813⊖	0.733⊖	0.840⊕	0.795⊖	0.838⊕
CC_3-10	0.748	0.845⊕	0.571⊖	0.618⊖	0.782⊕	0.899⊕	0.865⊕	0.837⊕
CC_11-30	0.791	0.873⊕	0.790⊖	0.785⊖	0.834⊕	0.907⊕	0.846⊕	0.850⊕
CC_31-100	0.863	0.896⊕	0.850⊖	0.851⊖	0.783⊖	0.887⊕	0.863	0.849⊖
CC_101-300	0.845	0.873⊕	0.851⊕	0.821⊖	0.750⊖	0.842⊖	0.808⊖	0.867⊕
MF_3-10	0.818	0.887⊕	0.630⊖	0.681⊖	0.850⊕	0.951⊕	0.880⊕	0.879⊕
MF_11-30	0.842	0.903⊕	0.861⊕	0.836⊖	0.865⊕	0.936⊕	0.884⊕	0.909⊕
MF_31-100	0.838	0.888⊕	0.892⊕	0.881⊕	0.843⊕	0.887⊕	0.884⊕	0.903⊕
MF_101-300	0.874	0.904⊕	0.894⊕	0.884⊕	0.843⊖	0.909⊕	0.844⊖	0.918⊕

For each of the 12 subsets, the of CLUS-HMC-ENS is compared with the MouseFunc systems. A win (⊕) means that the MouseFunc system outperforms CLUS-HMC-ENS, a loss (⊖) means that it is outperformed by CLUS-HMC-ENS.

ISSN: 1471-2105