Skip to main content
Figure 3 | BMC Bioinformatics

Figure 3

From: Comparison of methods for estimating the nucleotide substitution matrix

Figure 3

Results showing the subset of alignments on which the Lake method completed successfully. Results for probability of completion (left column) considering all matrices, and accuracy (right column) considering only the subset of matrices for which the Lake method completed, for the different methods (legend in panel a) under different simulation conditions. Completion is measured as the fraction of trials in which a given method completed; error is measured as the mean Euclidean distance between the matrix used to generate the sequences and the inferred matrix (arbitrary units). Figures show effects of DNA length (a, b), branch ratio (c, d) and branch inner length (e, f) on ability to complete successfully and accuracy respectively. Insets in panels b, d, and f show the subset of the data where the branch lengths were below 600 nucleotides, where the Goldman method performed relatively well: scale is the same as the main panel in each case, except for b (where the length ranges from 100 to 600 nucleotides, as shown). The effect of whether the inner branches evolved according to the same or different matrices was negligible, and is not shown.

Back to article page