Improving structure-based function prediction using molecular dynamics.

Area of Science:

• Structural biology
• Computational biology
• Biochemistry

Background:

• Increasing number of proteins with solved structures but unknown functions.
• Novel protein folds pose challenges for function prediction.
• Existing computational methods often rely on static structures, neglecting dynamics.

Purpose of the Study:

• To improve structure-based function prediction by incorporating molecular dynamics.
• To enhance the identification of functional sites, particularly calcium (Ca2+) binding sites.
• To evaluate the performance of dynamic structure-based prediction for challenging proteins.

Main Methods:

• Coupling molecular dynamics (MD) simulations with structure-based function prediction algorithms.
• Focusing on the identification of calcium (Ca2+) binding sites.
• Applying the integrated approach to 11 challenging proteins with unknown functions.

Main Results:

• Substantial improvement in the performance of function prediction.
• Identification of additional functional sites, with up to 22 more sites found in one case.
• A modest increase in apparent false positives was observed.

Conclusions:

• Treating molecules as dynamic entities significantly enhances structure-based function prediction.
• Molecular dynamics simulations provide valuable insights into protein function.
• This approach offers a more effective strategy for annotating protein functions, especially for novel folds.