Skip to main content
Figure 7 | BMC Bioinformatics

Figure 7

From: Selecting informative subsets of sparse supermatrices increases the chance to find correct trees

Figure 7

Summary of results of tree reconstructions of the four different set ups. In this figure, columns (org) refer to tree reconstructions based on unreduced matrices, columns (w) refer to tree reconstructions based on reduced weighted matrices B, columns (uw) refer to tree reconstructions based on reduced unweighted matrices B, columns (ca, cb) refer to tree reconstructions based on the application of thresholds to genes only and columns (cc, cd) refer to tree reconstructions based on the application of thresholds to genes and taxa concerning data coverage. Set1 refers to datasets with relatively low among gene rate heterogeneity and set 2 refers to datasets with relatively high among gene rate heterogeneity. Columns labeled (gaussian) illustrate results derived from datasets with gaussian distribution of missing data, columns labeled (power-law) illustrate results derived from datasets with power-law and non-random distribution of missing data and columns labeled (thresholds) illustrate results derived from datasets reduced by applied predefined thresholds of data coverage. In (A) the distribution of d QD values for each simulation and set up is presented as box plots with median (black bar), quartiles (box), whiskers (doted range) and outliers (circles). The smaller d QD , the more similar is the reconstructed tree to the tree used in data simulations. In (B) columns represent the frequency of correct tree reconstruction for each set up.

Back to article page