Figure 5.

Influence of outliers on false negative and false positive error rates on a sample size of m = 20. In a similar manner to figure (3), simulated data were imposed by technical errors, but in addition, by the presence of outliers that were located in a 2–1000 σ distance from the linear function. The algorithm proved to be robust against such outliers.

Kose et al. BMC Bioinformatics 2007 8:162   doi:10.1186/1471-2105-8-162
Download authors' original image