Table 2.
Effects of 2010 rotamer library smoothing in SCWRL4 and Rosetta
| 2002 | Optim | 2%↓ | 5%↓ | 10%↓ | 20%↓ | 25%↓ | 2009it10 | ||
|---|---|---|---|---|---|---|---|---|---|
| Side-chain prediction | |||||||||
| SCWRL4 | D(’10,’02), asymm. ED25–100% | 83.04% | +0.57% | +0.61% | +0.67% | +0.40% | +0.11% | −0.08% | N.D. |
| SCWRL4 | D(’10,’02), symm., ED0–100% | 79.33% | +0.55% | +0.59% | +0.64% | +0.39% | +0.11% | −0.11% | N.D. |
| Rosetta FastRelax | D(’10,’02), symm., ED0–100% | 72.82% | +0.48% | +0.21% | +0.10% | −0.09% | −1.04% | −1.44% | −1.45% |
| Rosetta ClassicRelax | D(’10,’02), symm., ED0–100% | 76.12% | +0.21% | +0.13% | 0.00% | −0.12% | −0.81% | −1.12% | −1.57% |
| RMSD differences | |||||||||
| Rosetta FastRelax: | D(’10,’02) / ’02 (backbone) | 1.112 | −2.23% | −0.37% | −0.19% | +0.04% | +0.67% | +1.36% | +0.63% |
| D(’10,’02) / ’02 (all atoms) | 1.596 | −1.83% | −0.49% | −0.58% | −0.19% | +0.76% | +1.22% | +1.01% | |
| Rosetta ClassicRelax: | D(’10,’02) / ’02 (backbone) | 1.081 | −1.21% | −0.67% | −2.06% | −1.35% | +0.64% | +2.41% | +0.18% |
| D(’10,’02) / ’02 (all atoms) | 1.517 | −0.02% | +0.28% | −0.76% | −0.21% | +1.36% | +2.36% | +1.91% | |
| TotalScoreMinusDun | |||||||||
| Rosetta FastRelax: | D(’10,’02) | −382.28 | 1.783 | −0.004 | −1.035 | −2.436 | −4.884 | −5.645 | −1.965 |
| Rosetta ClassicRelax: | D(’10,’02) | −379.00 | 0.871 | −0.692 | −1.548 | −2.685 | −5.068 | −5.504 | −1.970 |
2010 library names are listed in the first row. 2009it10 is a modified version of a developmental rotamer library created by using similar methods (with some important differences) in 2008. It is distributed with Rosetta3 and was recently described by Song et al. (2011). For side-chain accuracy, the absolute average percentage accuracy is given for the 2002 library, and the differences from those values are given for the other libraries (2010Library – 2002). For RMSD differences, the mean RMSD in (Å) is given for the 2002 library, and the percent differences from 2002 are given for the 2010 libraries. For TotalScoreMinusDun, the mean values are given for the 2002 library, and the differences (in Rosetta score units) are given for the 2010 libraries. For side-chain accuracy, “symm” indicates Asn, His, and Gln terminal dihedrals were treated as symmetric, while “asymm” indicates they are treated like other dihedral angles. ED25–100% indicates only side chains with electron density in the 25th to 100th percentile were included in the accuracy assessment. ED0–100% means all side chains were included. Better numbers are in bold type (higher side-chain accuracy, lower RMSDs), while worse numbers are in italic type. N.D.=not done. See also Figure S3.