Figure 5
From: Learning sparse models for a dynamic Bayesian network classifier of protein secondary structure
The relationship between model parameters and secondary structure element types. (A) The figure plots, for helices, the learned non-zero dlink parameters at the 90% sparsity level as a function of the offset from the current amino acid. The mean of positive (red squares) and negative values (black squares) are displayed separately. (B)-(C) The figures show similar plots for β strands and loops, respectively. Certain positions retain more dlink parameters than the others demonstrating the correlation behavior between a given position and its local neighbors.