Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: Metabolomics variable selection and classification in the presence of observations below the detection limit using an extension of ERp

Fig. 2

Null CDF for \( {\widehat{er}}_{up\;}^{*} \) with \( \pi \) taking on different values. One million simulation repetitions, group sizes \( {N}_0={N}_1=20 \) and the weight set \( {w}_0={w}_1=\frac{1}{2} \) were used to calculate the null-distributions. Each line represents the CDF for a different value of \( \pi \), \( \pi =0 \) (black), \( \pi =0.2 \) (light blue), \( \pi =0.4 \) (purple), \( \pi =0.6 \) (dark blue) and \( \pi =0.8 \) (red). The CDFs are plotted on a log10 scale for clarity purposes, since interest centres on the extreme left tail

Back to article page