Figure 2.

Performance comparisons from real ChIP-chip datasets. The upper panel of the figure shows the box plots of the distribution of motif similarity scores across six different weight prior configurations. The lower panel of the figure shows the box plots of the distribution of CPU hours used by six prior assumptions over the weights. Here, the red line represents the prior assumptions over the weights without inclusion of the nucleosome information, and the blue line indicates the prior assumptions over the weights with the inclusion of the nucleosome information. G4, L4, and C4 indicates Gaussian prior, Laplace prior, and Cauchy prior assumptions to the weights with four classes of regularization constants α, respectively.