Skip to main content

Table 1 ROC values

From: Considering scores between unrelated proteins in the search database improves profile comparison

Scheme ROC(2 104) ROC(2 105) ROC(745550)
Query × Cal 0.1020 ± 2 10-4 0.2099 ± 7 10-5 0.3135 ± 3 10-5
Query × Cal, Subj × DB 0.1427 ± 2 10-4 0.2819 ± 7 10-5 0.3990 ± 3 10-5
Shuffled DB, SS = C 0.1093 ± 2 10-4 0.2444 ± 7 10-5 0.3718 ± 3 10-5
Shuffled DB, original SS 0.1248 ± 2 10-4 0.2688 ± 7 10-5 0.3983 ± 3 10-5
Query, Subj × Cal 0.1146 ± 2 10-4 0.2382 ± 7 10-5 0.3540 ± 3 10-5
Subj × Cal 0.1074 ± 2 10-4 0.2268 ± 7 10-5 0.3462 ± 3 10-5
Subj × DB 01437 ± 2 10-4 0.2808 ± 7 10-5 0.3959 ± 4 10-5
Query × Cal × 4, Subj × DB 0.1323 ± 2 10-4 0.2676 ± 7 10-5 0.3844 ± 3 10-5
Query × Cal × 2, Subj × DB 0.1384 ± 2 10-4 0.2767 ± 7 10-5 0.3944 ± 3 10-5
Query × DB 0.1417 ± 2 10-4 0.2690 ± 7 10-5 0.3711 ± 3 10-5
Query × DB, Subj × DB 0.1607 ± 2 10-4 0.3052 ± 7 10-5 0.4184 ± 3 10-5
  1. Receiver operating characteristics (ROC) for tested statistical schemes, calculated for different numbers of top false positives: the mean of 5 top false positives per query (2 104 false positives total); the mean of 50 top false positives per query (2 105 false positives total), and the point where PROCAIN retrieves half of all true positives in the dataset (745550 false positives total). The total number of true positives in the testing set is T = 474929. Statistical schemes are denoted the same way as in Fig. 1-4.