This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Experiments selected by the algorithm improve GBA-based function prediction.
ROC curve analysis for evaluating the effectiveness of the selected experiments at improving GBA-based gene function prediction. The functional categories shown here are the same as in Figure 5, while more examples for both the GO and MIPS functional categorizations are presented in Supplementary Information S7. (A) Average ROC curves from 10-fold cross validation obtained using all experiments (red) and the experiments selected by our algorithm (green) for four different GO functional categories, two from Arabidopsis and two from yeast. (B) Averaged (1-AUC) scores from 10-fold cross validation – the lower the value of (1-AUC), the better the performance of the classifier. (C) Average (1-AUC) and average ROC curves for two MIPS FunCat terms for Arabidopsis. p-values for the t test between the ten (1-AUC) values from the ten-fold cross-validation obtained using the selected experiments and those obtained using all experiments are also reported in blue for both (B) and (C).