Table 2.
Comparison of the performance between IQP and GMFIT fitting
| Assembly | Comp. | GMFIT |
IQP |
|||
|---|---|---|---|---|---|---|
| Time (s) | RMSD | Time (s) | RMSD | RMSD* | ||
| 1AFW | 2 | 7.1 | 1.0 | 0.15 | 4.8 | 0.9 |
| 1NIC | 3 | 16.0 | 1.8 | 0.65 | 5.9 | 1.1 |
| 2REC | 6 | 110.9 | 2.3 | 2.56 | 1.7 | 1.0 |
RMS errors between the fitted and native structures are shown for three different assemblies. GMFIT used 16 GDFs (Gaussian distribution functions) to represent the atomic structures of each component and 12 GDFs to represent the density map of the assembly (Kawabata, 2008). The IQP model used 5 feature points per component for each assembly. The individual columns are: Assembly, the PDB ID (Bernstein et al., 1977) of the assembly structure being used; Comp., the number of components of the assembly; RMSD, RMS error between the corresponding Cα atoms in the fitted and assembly structures; RMSD*, the RMS error after refinement.