Skip to main content
Figure 1 | BMC Structural Biology

Figure 1

From: Computational identification of residues that modulate voltage sensitivity of voltage-gated potassium channels

Figure 1

Flow chart of procedures followed to develop the optimal V 50 predictor. The data set was subjected to several different learning algorithms, either alone or in combination with two types of feature selection. The KNN learning algorithm and the wrapper feature selection algorithm (highlighted in red) were found to yield the best results. These algorithms were then used to evaluate the effect of removing sequences as outliers to yield the data set used for construction of the final predictor. The individual processes that were used to construct the final predictor are highlighted in red.

Back to article page