Skip to main content
. 2025 Jun 3;10(23):24030–24049. doi: 10.1021/acsomega.4c08181

3. MolProbity Data Obtained for Models Generated from AFv2, AFv3, and the Hybrid Homology Modeling (HHM) Method that Used High-Confidence AFv2 Mouse ORs as Templates .

ORs model MolProbity score Cβ-Dev. [goal:0] bad bonds [goal:<0.1%] bad angles [goal:<0.1%] CaBLAM out- liers [goal:<1.0%]
OR51E2 Exp. 1.82 0 0/2557 0/3302 2 (0.7%)
  AFv2 1.15 0 0/2557 6/3478 (0.17%) 1 (0.3%)
  AFv3 0.58 0 0/2638 8/3589 (0.22%) 2 (0.66%)
  AF2-Hyb 0.83 0 0/2557 3/3478 (0.09%) 1 (0.3%)
  AF3-Hyb 0.70 0 0/2528 5/3441 (0.15%) 0
OR51E1 AFv2 0.96 1 (0.33%) 0/2539 0.23 0
  AFv3 0.96 1 (0.33%) 0/2539 6/3469 (0.17%) 2 (0.66%)
  AF2-Hyb 0.99 0 0/2533 3/3450 (0.09%) 1 (0.3%)
OR51D1 AFv2 1.42 1 (0.33%) 0/2580 5/3514 (0.14%) 1 (0.3%)
  AFv3 1.58 0 0/2580 9/3514 (0.26%) 3 (0.9%)
  AF2-Hyb 0.75 0 0/2554 4/3480 (0.11%) 0
OR51G2 AFv2 1.57 1 (0.33%) 0/2520 6/3423 (0.18%) 2 (0.6%)
  AFv3 1.46 0 0/2520 3/3423 (0.09%) 3 (1.9%)
  AF2-Hyb 0.70 0 2/2514 (0.08%) 3/3414 (0.09%) 1 (0.3%)
a

It can be observed that for almost all four receptors, the HHM method has generated a better MolProbity score. The score encompasses Clashscore, Rotamer, and Ramachandran evaluations in a single score, which then, is then normalized to be on the same scale as X-ray resolution.