Comparative analysis and predictive performance of CSR transcripts in PAM50 breast cancer subtypes.

(A) Bar graphs showing the number of differentially expressed CSR transcripts between each pairwise comparison of the PAM50 breast cancer subtype (x-axis). The two bar graphs are plotted for the upregulated transcripts between each comparison (left) and downregulated transcripts between each comparison (right; also see S3 File). (B) A plot of the confusion matrix. The diagonal cells correspond to correctly classified observations (green cells). The off-diagonal cells correspond to incorrectly classified observations (red cells). The number of observations and the percentage of the total observations are shown in each cell. The column to the far right shows the precision (or positive predictive value rate) in green text and the false discover rate shown in red text. The row at the bottom of the plot shows the recall (or true-positive rate) in green text and the false-negative rate in red. The cell in the bottom right of the plot shows the overall accuracy. (C) shows the ROC-AUC (Receiver Operated Characteristic-Area Under the Curve) for the PAM50 breast cancer classes. The ROC curves show the true-positive rate versus the false-positive rate for predictions of the classes made by the trained classifier. The coloured markers on the plots show the values of the false positive rate and the true-positive rate of the trained classifier toward predicting the PAM50 subtype of breast cancer using CSRs.