Skip to main content
Fig. 5 | Journal of Cheminformatics

Fig. 5

From: Off-targetP ML: an open source machine learning framework for off-target panel safety assessment of small molecules

Fig. 5

Box plots comparing the overall balanced accuracy for high and low hit percent target groups, with respect to t the a Machine learning methods and b Target protein classes. The five methods are represented on the x axis and the balanced accuracy values on the y axis. The red box plots represent the high hit percent target groups and the blue box plots represent the low hit percent target groups. The High hit percent target groups achieve higher balanced accuracies irrespective of the method used. The target classes are represented on the x axis and the balanced accuracy values of the neural networks on the y axis. For some target classes, no significant difference is seen in the overall balanced accuracy between the high hit percent and low hit percent groups (e.g. GPCR class) while for other classes (e.g. Ion-channels), a significant difference is seen in the overall balanced accuracy between the high and low hit percent groups. The hit percent is represented on the x axis and the balanced accuracy on the y axis. Each circle represents a target. The circles are color coded according to the target classes. The size of the circles varies according to the dataset size

Back to article page