Fig. 5

Correlation circle plot of sequence and position classification scores for Lrp targets. The two axis represent the first two “variates” of CCA (i.e. the two components which capture the highest correlation between variables). Here we plot the projection of each correlation score on each variate for the selected boosting position classifiers (red) and the selected sequence classifiers (blue). The correlation between two points is negative if the angle that connects them (with the origin as the angle vertex) is obtuse and positive if the angle is sharp [32]. For the Lrp target prediction the selected boosting classifiers (apart from the pair of position classifier No5 and sequence classifier No2) are connected by obtuse angle with the origin as the vertex, indicating negative correlation between position and sequence scores. (data points are named as follows <classifier name><TF name><boosting iteration>)