TABLE 1.
Predictive performance of the statistical fitness when different groups of homologous sequences are used in the starting MSA
| Designation | Length (no. of aa)a | No. of sequences | Effective no. of sequencesb (±SD) | AUC (±SE)c |
|
|---|---|---|---|---|---|
| Epistatic model | Independent model | ||||
| Alllax-29k | 141–199 | 29,498 | 6,352 | 0.894 | 0.807 |
| Allstringent-23k | 155–186 | 23,176 | 4,809 | 0.856 | 0.857 |
| Alllax-3k | 3,037 | 1,420 ± 33 | 0.815 ± 0.013 | 0.744 ± 0.026 | |
| Bacterialax-27k | 137–200 | 26,950 | 5,565 | 0.868 | 0.888 |
| Bacteriastringent-23k | 149–188 | 23,194 | 4,595 | 0.851 | 0.871 |
| Bacterialax-3k | 3,037 | 1,309 ± 39 | 0.839 ± 0.006 | 0.782 ± 0.013 | |
| Firmicuteslax-3k | 133–201 | 3,037 | 940 | 0.852 | 0.795 |
| Firmicutesstringent-2k | 163–192 | 2,007 | 600 | 0.840 | 0.830 |
| Firmicutes+alllax-3k | 3,037 | 1,344 ± 34 | 0.856 ± 0.005 | 0.818 ± 0.014 | |
| Firmicutes+Bacterialax-3k | 3,037 | 1,317 ± 30 | 0.851 ± 0.004 | 0.814 ± 0.014 | |
| Firmicutes+nonbacterialax-2k | 5,585 | 0.865 | 0.799 | ||
a
The acceptable amino acid lengths across the six initial groupings.
b
The effective number is the sum of the inverse of the neighborhood size of each sequence, where the neighborhood is defined as the number of sequences within 80% identity. SD, standard deviation.
c
Results are presented as mean values from 20 subsamplings from the parent group(s). SE, standard error.