Figure 6

Bayesian maximum likelihood tree of training data and independent test data. The set of 54 VKC sequences used for the final predictor and the 13 VKC sequences from the independent test set were aligned using MUSCLE [70] and the alignment was trimmed to 296 residues. This data set was used to generate a maximum likelihood tree using MrBayes [41], using a total of 500,000 cycles with a 250 cycle burn-in. The training set channels are labelled with their VKCDB identification number and their V₅₀ value, the 13 test sequences are labelled with the name of the species and the channel, and were highlighted in red (Figure 6).