Summary of GSFE-refinement and other refinement methods on the 150-target refineD dataset (results for other methods are taken from ref. 22)a.
| Method | Avg. top 1 | Avg. best of 5 | GDT-HA-num |
|---|---|---|---|
| refineD-C | 0.6365 | 1.3109 | 121/150 |
| refineD-NC | −1.2403 | 1.5343 | 104/150 |
| FG-MD | 0.5597 | 0.5597 | — |
| FastRelax | −3.4317 | −0.1999 | — |
| FastRelax-0.5 Å | −0.3411 | 0.8811 | 90/150 |
| FastRelax-2.0 Å | −1.2120 | 0.8223 | 77/150 |
| FastRelax-4.0 Å | −2.5471 | 0.0751 | 67/150 |
| ModRefiner-0 | −0.8400 | −0.8400 | — |
| ModRefiner-100 | 0.1491 | 0.1491 | — |
| GSFE-refinement | 0.0800 | 0.4400 | 112/150 |
a
More details for top 1 and the best of top 5 models for refineD data set are available in ESI (Tables S1 and S2).