Figure 8

Network-based prediction of quantitative cell morphology phenotypes. A wide variety of phenotypes based upon quantitative yeast cell shape and intracellular features [46] are predictable, as shown for the ten phenotypes in this receiver operating characteristic (ROC) analysis (selected from S. cerevisiae Morphology Database [SCMD] phenotypes with area under the ROC curve [AUC] > 0.68). For each of the features, the 40 genes whose deletion mutants show either the 40 highest or 40 lowest values for that quantitative feature (indicated by 'high' or 'low', respectively) were selected as the seed gene set. Predictability was evaluated using ROC analysis as in Figure 2, plotting the true positive prediction rate versus false positive rate, using leave-one-out cross-validation. For clarity, the line connecting the final point of each graph to the top right corner has been omitted. Labels of features are adapted for clarity from the SCMD [50]; the SCMD labels A, A1B, and C represent unbudded cells, budded cell with one nucleus in mother cell, and large-budded post-mitotic cells with nuclei in both mother and daughter cell, respectively. Ratio measurements refer to proportions across a population of cells. FN, false negative; FP, false positive; TN, true negative; TP, true positive.