Comparison of error rates in identification of microtubule ends in COS cells.

Microtubule ends identified by the dCCD method or the plusTipTracker program are compared with ground-truth data (GTD) based on human identification in raw images (as illustrated in Figure S3). False positive rates are calculated as (1-O/(O+P), while false negative rates as (1-O/(O+N)), where O is the number of ends found by the dCCD method and GTD, P is the number only found by the dCCD method, and N is for those missed by the dCCD method. Data expressed in mean±s.e were pooled from randomly selected regions (∼8% of the total imaging area per frame, with 2160 ends examined) of 10–11 frames from 3 COS cells.