## Table 1 |
|||

Prediction of tissue malignancy by metabolite ratios and by nearest centroid classification |
|||

method |
# of metabolites |
sensitivity |
specificity |

metabolite ratios: CMP / pentadecanoic acid | 2 | 94.8% | 93.9% |

metabolite ratios: majority vote of 91 classifiers | 20 | 97.0% | 93.9% |

cancerclass: nearest centroid | 2 | 95.1% (86.4% - 97.6%) | 93.9% (81.8% - 100%) |

cancerclass: nearest centroid | 20 | 97.1% (94.2% - 99.1%) | 93.9% (84.8% - 100%) |

cancerclass: nearest centroid | 162 | 95.1% (93.2% - 97.1%) | 93.9% (84.8% - 100%) |

CMP : cytidine-5-monophosphate.The metabolite ratio method was developed in the current work. Nearest centroid classification was executed using the R package cancerclass [37]. For the latter method, classification rates including 95% confidence intervals were estimated using multiple random validation.

Budczies * et al.*

Budczies * et al.* *BMC Genomics* 2012 **13**:334 doi:10.1186/1471-2164-13-334