Learning probabilistic models of hydrogen bond stability from molecular dynamics simulation trajectories
2011

Learning Models of Hydrogen Bond Stability

Sample size: 4000 publication Evidence: moderate

Author Information

Author(s): Chikalov Igor, Yao Peggy, Moshkov Mikhail, Latombe Jean-Claude

Primary Institution: King Abdullah University of Science and Technology

Hypothesis

Can probabilistic models predict hydrogen bond stability better than energy-based models?

Conclusion

The study shows that machine learning models can predict hydrogen bond stability more accurately than traditional energy-based methods.

Supporting Evidence

  • The models built can predict hydrogen bond stability significantly better than energy-based models.
  • About 80% of the least stable hydrogen bonds predicted were actually among the least stable in reality.
  • The study used data from six molecular dynamics simulation trajectories.

Takeaway

This study teaches us how to use computer models to understand how strong hydrogen bonds are in proteins, which helps us know how proteins work.

Methodology

Inductive learning methods were used to train probabilistic models based on molecular dynamics simulation data.

Limitations

The models may not perform well with short MD simulation trajectories and could benefit from more diverse data.

Digital Object Identifier (DOI)

10.1186/1471-2105-12-S1-S34

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication