Fig. 2From: Grouping of nanomaterials to read-across hazard endpoints: from data collection to assessment of the grouping hypothesis by application of chemoinformatic techniquesHierarchical clustering of the TiO2 analogues. The numbers in red correspond to the “Approximately Unbiased” (AU) p-value that is computed by multiscale bootstrap resampling, and the ones in green to “Bootstrap Probability” p-value (BP), which is computed by normal bootstrap resampling. The height in the Y-axis indicates the distance between clusters computed as average linkage. AU p-value will be used for the interpretation as it is usually a better approximation to the real p-valueBack to article page