Figure 2

Analysis of scores of subject to all database non-homologs has a dominant effect compared to query calibration. The performance of the query calibration alone (black) and combined with subject calibration on the full database (red), compared to the subject calibration alone, using either the calibration database (green) or the full set of non-homologs (blue). Surprisingly, the latter scheme provides virtually the same detection quality regardless of whether query calibration scores are additionally considered (blue Vs red curve).