Figure 3.

Type I error rates. Type I error rates, for the six methods providing nominal p-values, in simulation studies <a onClick="popup('http://www.biomedcentral.com/1471-2105/14/91/mathml/M22','MathML',630,470);return false;" target="_blank" href="http://www.biomedcentral.com/1471-2105/14/91/mathml/M22">View MathML</a> (panel A), <a onClick="popup('http://www.biomedcentral.com/1471-2105/14/91/mathml/M23','MathML',630,470);return false;" target="_blank" href="http://www.biomedcentral.com/1471-2105/14/91/mathml/M23">View MathML</a> (panel B), <a onClick="popup('http://www.biomedcentral.com/1471-2105/14/91/mathml/M24','MathML',630,470);return false;" target="_blank" href="http://www.biomedcentral.com/1471-2105/14/91/mathml/M24">View MathML</a> (panel C) and <a onClick="popup('http://www.biomedcentral.com/1471-2105/14/91/mathml/M25','MathML',630,470);return false;" target="_blank" href="http://www.biomedcentral.com/1471-2105/14/91/mathml/M25">View MathML</a> (panel D). Letting some counts follow a Poisson distribution (panel B) reduced the type I error rates for TSPM slightly but had overall a small effect. Including outliers with abnormally high counts (panels C and D) had a detrimental effect on the ability to control the type I error for edgeR and NBPSeq, while DESeq became slightly more conservative.

Soneson and Delorenzi BMC Bioinformatics 2013 14:91   doi:10.1186/1471-2105-14-91
Download authors' original image