"Conditional Restraints": Restraining the Free Atoms in ARP/wARP

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Automated protein model building into electron density maps is a complex challenge in structural biology.
• Current methods like ARP/wARP often discard valuable structural information from chemically unassigned atoms.
• This loss of information hinders the accuracy and efficiency of automated model building processes.

Purpose of the Study:

• To develop and implement a novel approach for improving automated protein model building.
• To address the limitations of discarding structural information from chemically unassigned atoms.
• To enhance the accuracy of protein model building by incorporating geometric and tentative assignment-based restraints.

Main Methods:

• Applied restraints between free atoms and between free atoms and partial protein models.
• Utilized geometric considerations of protein structure and tentative (conditional) assignments for free atoms.
• Integrated these restraints into the REFMAC5 refinement program, allowing for dynamic, step-by-step adjustments.

Main Results:

• Demonstrated significant improvements in automated building for individual structures across a large dataset.
• Showcased enhanced accuracy in protein model building using experimentally phased and molecular replacement structures.
• Validated the effectiveness of conditional restraints in overcoming limitations of existing automated methods.

Conclusions:

• The developed concept and implementation of conditional restraints drastically improve automated protein model building.
• This approach effectively leverages geometric information and tentative assignments, preserving crucial structural data.
• The method shows potential for broader applications, including restraining geometries like hydrogen bonds in low-resolution refinement.