Figure 7.

Cross-validation testing. The average fraction of correctly identified coiled-coil interactions as a function of the number of FP. Solid lines give averages computed using the appropriate cross-validation weights, and dotted lines give averages computed using the base-optimized weights. The TP rates shown were averaged over all human bZIP sequences, shown in purple, over human bZIP sequences with CV-similarity equal to or greater than 50%, shown in red, and over human bZIP sequences with CV-similarity less than 50%, shown in blue.