Table 4.
The performance metrics of the three programs on a dataset that uses domain definitions derived from the CATH database
| Method | Number of domains |
|||
|---|---|---|---|---|
|
One |
Multi |
|||
| A | Sp | Sn | A | |
| FIEFDom | 92 | 91 | 65 | 61 |
| PPRODO | 90 | 58 | 51 | 37 |
| DOMpro | 91 | 58 | 21 | 18 |
| FIEFDom (only two domains) | 89 | 91 | 50 | 48 |
| FIEFDom (only multidomain) | 89 | 91 | 62 | 58 |
All values are percentages. Five prediction sets were generated to understand how FIEFDom (with three versions of the same RPS), PPRODO and DOMpro perform on a database that derives its domain definitions from the CATH database (version 2.5.1). The results for two-, three- and four-domain proteins have been averaged and are shown under ‘Multi’. The first row shows the performance of FIEFDom that uses the SCOP 1.65 (30%) database as the RPS. The second and third rows show the performance of PPRODO and DOMpro, respectively. The fourth and the fifth rows show the performance of FIEFDom that uses a RPS containing only two-domain proteins or multidomain proteins, respectively.