Figure 2

Average cumulative number of false positives versus the effective variable to 1 giga residues.

The effective variable is E-value for search methods that report E-values, and becomes e^{-(Quality score)} for others. Except for X!Tandem and RAId_DbS, the effective variables of other methods do have a database size dependence and size calibrations are needed. Except for RAId_DbS, all other methods need calibration transformations to reach the calibrated E-value. Theoretically speaking, average number of false positives with E-values less than or equal to a cutoff E _c should be E _c provided that the number of trials is large enough. And this is how our E-value calibration is done. In the last panel, we plot the calibrated E-value as the abscissa and the average number of false positives with that E-value or smaller as the ordinate. The calibrations are able to bring the calibrated E-values close to the average number of false positives as theoretically expected.