Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: MOST: most-similar ligand based approach to target prediction

Fig. 2

Prediction results of MOST in one dataset of sevenfold cross-validation with Logistic Regression method and Morgan fingerprint. a and b, the predicted results derived from different “active” data definition: pKi ≥ 5 and pKi ≥ 6. Results generated by using explicit and implicit Ki of most-similar ligand in model training are compared. Left panels, the predicted results in Tc most vs pKi most scatter plot. Middle panels, the fraction of data regarding to the increasing threshold of Tc. Right panels, the data fraction regarding to the decreasing threshold of p values. The difference between fTP and fFP was plotted in black, dash line. In all panels, true positives are colored red, while true negatives are blue; false positives are cyan, while false negatives are orange

Back to article page