Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: Inverse similarity and reliable negative samples for drug side-effect prediction

Fig. 4

Differences among similarity measurements and similarity integration methods. In each panel, the x-axis denotes the index of each side-effect and the y-axis denotes the F1-score difference between two methods. For instance, Fig. 4a describes the differences between “ComMean” and “ComGM” using the F1-score of each side-effect from ComMean minus that from ComGM (i.e. difference= F1-score(ComMean) −F1-score(ComGM)). Thus we can identify which method performs better by comparing the area under the curve above zero (i.e., area A) with the area above the curve under zero (i.e., area B). Panel (a) shows the F1-score difference between “ComMean” and “ComGM”; Panel (b), (c) and (d) illustrate the F1-score difference between “ComMax” and “ComGM/ComMean/ChemTarMax” respectively; Panel (e) and (f) shows the F1-score difference between “ChemTarMax” and “Chem/Tar” respectively

Back to article page