Table 1.
Summary of mean RMSDs (Å) for the different test sets.
| Nanobody test set | |||||||
|---|---|---|---|---|---|---|---|
| Method | VHH | Fr | CDR1 | CDR2 | CDR3 | ||
| RosettaAntibody | 2.68±1.12 | 1.56±0.94 | 3.41±1.61 | 2.40±1.39 | 6.38±2.59 | ||
| AlphaFold2 | 1.51±0.64 | 0.93±0.42 | 2.35±1.16 | 1.51±0.95 | 2.88±2.04 | ||
| AlphaFold2-No MSA | 10.37±4.78 | 10.06±5.01 | 11.30±7.71 | 8.02±5.15 | 11.06±4.51 | ||
| NanoNet+Modeller | 1.65±0.53 | 1.02±0.43 | 2.72±1.22 | 1.72±0.95 | 3.20±1.44 | ||
| NanoNet+SCWRL | 1.60±0.52 | 0.99±0.44 | 2.66±1.15 | 1.68±0.95 | 3.09±1.39 | ||
| mAb test set | |||||||
| Method | Fr | CDR1 | CDR2 | CDR3 | |||
| RosettaAntibody | 1.25±0.62 | 1.42±0.98 | 1.66±1.83 | 6.56±3.29 | |||
| DeepAb | 0.43±0.18 | 0.72±0.66 | 0.85±0.81 | 2.33±1.32 | |||
| NanoNet+Modeller | 0.64±0.18 | 0.88±0.67 | 0.95±0.82 | 2.38±1.17 | |||
| NanoNet+SCWRL | 0.60±0.19 | 0.83±0.65 | 0.89±0.85 | 2.29±1.09 | |||
| TCR test set | |||||||
| Method | V β | Fr | CDR1 | CDR2 | CDR3 | ||
| Rosetta TCRmodel | 1.49±0.68 | 1.07±0.60 | 0.93±0.33 | 1.24±1.29 | 2.79±1.28 | ||
| NanoNet+Modeller | 1.31±0.52 | 1.04±0.43 | 0.91±0.49 | 1.27±1.23 | 2.18±0.81 | ||
| NanoNet+SCWRL | 1.28±0.48 | 1.01±0.39 | 0.93±0.49 | 1.27±1.19 | 2.12±0.76 | ||
| AlphaFold2* | 1.48±0.58 | 0.89±0.19 | 0.99±0.61 | 1.19±0.63 | 3.07±1.45 | ||
| AlphaFold2-No MSA* | 15.22±5.91 | 15.54±6.15 | 18.71±9.73 | 12.57±4.67 | 11.10±5.97 | ||
| NanoNet+Modeller* | 1.72±0.58 | 1.24±0.43 | 1.15±0.74 | 2.12±1.94 | 2.78±0.83 | ||
| NanoNet+SCWRL* | 1.68±0.53 | 1.20±0.38 | 1.19±0.74 | 2.11±1.86 | 2.71±0.81 | ||
Nb test set (37 Nbs), mAb test set (47 mAbs), and TCR test set (15 TCRs). For Rosetta TCRmodel only 14 TCRs could be modeled. For AlphaFold2 only 5 TCRs were not used for training. The results for these 5 TCRs are indicated with*.