Fig. 4: Performance evaluation on the MoA prediction task. | Nature Communications

Fig. 4: Performance evaluation on the MoA prediction task.

From: DTIAM: a unified framework for predicting drug-target interactions, binding affinities and drug mechanisms

Fig. 4

A–D Performance comparison of DTIAM with AI-DTI on the Activation and Inhibition datasets under three experiment settings in terms of AUPR and AUROC. All results were obtained by 5-fold cross-validation. The ratio between the positive and negative samples is 1:10. Box plots show the median as the center lines, upper and lower quartiles as box limits, whiskers as maximum and minimum values, and dots represent outliers. E–H Validating DTIAM on 10 specific targets. E Prediction accuracy for each target, with the score at the top of the bar and the ratio inside indicating the number of correctly predicted drugs out of the total number. F–G The recall of activation and inhibition, respectively, with the top value indicating recall and the ratio inside indicating the number of correctly predicted activators or inhibitors out of the total number. H AUROC score for each target.

Back to article page