Assessment of Gaussian radial basis function network on protein secondary structures

Studies of the radial basis function (RBF) network on protein secondary structures are presented. Secondary structure prediction is a useful first step in understanding how the amino acid sequence of protein determines the native-state. If the secondary structure is known, it is possible to derive a comparatively small number of tertiary structures using the secondary structural element pack. A study of the Gaussian-RBF with different window sizes on the dataset developed by Qian-Sejnowski, and also a dissimilar dataset by Chandonia is given. The RBF network predicts each position in turn-based on a local window of residues, by sliding this window along the length of the sequence. It is shown that the Gaussian RBF network is not an appropriate technique to be used in the prediction of secondary structure for sequence structural state.