Figure 9
From: Predicting gene function in a hierarchical context with an ensemble of classifiers
Petite assay on the oms1Δgutant. To further evaluate the performance of our ensemble algorithm, we applied our methodology to S. cerevisiae. We tested the second most significant uncharacterized gene with predicted involvement in mitochondria organization and biogenesis (GO:0007005). For this GO term, the Bayesian integration of single-dataset SVM was selected as the best performer in the ensemble. Using an assay for measuring the frequency of petite colonies for deletion mutants, we found that compared to (a) the wild type, (b) the oms1Δ mutant showed 67% more cells without functional mitochondria. GO, Gene Ontology; SVM, support vector machine.