Figure 1

Correlation between the over-representation and the background rareness. ORS _r scores for the functional models M _t , the random models $M_{r_q}$, and the conserved models $M_{c_q}$ for q = 1, 2, 3, . . ., 1000 are plotted for each dataset with 200bp and 500bp promoters in the left and in the right column, respectively. Threshold θ = R(M _t ) + std(M _t )λ, λ = 0.0 is used. The figure depicts a rareness interpretable visualization through the statistical over-representation property of the functional motifs by showing that, the ORS _r (M _r ) scores are found distant from the ORS _r (M _t ) scores for all cases which implies that the random models have close to zero chance of being over-represented with comparison to the true models. In addition to this, the ORS _r (M _t ) scores are found to be mostly rare with comparison to the ORS _r (M _c ) scores, i.e., these non-functional conserved models have a rare chance of having better over-representation scores than the true models, for most of the datasets.