Inference of gene regulatory network from expression data is a challengingtask. Many methods have been developed to this purpose but a comprehensiveevaluation that covers unsupervised, semi-supervised and supervised methods,and provides guidelines for their practical application, is lacking. We performed an extensive evaluation of inference methods on simulatedexpression data. The results reveal very low prediction accuracies forunsupervised techniques with the notable exception of the z-score method onknock-out data. In all other cases the supervised approach achieved the highestaccuracies and even in a semi-supervised setting with small numbers of onlypositive samples, outperformed the unsupervised techniques.
展开▼