Entropy-based distance cutoff for protein internal contact networks

Area of Science:

• Biophysics
• Computational Biology
• Structural Biology

Background:

• Protein structure networks (PSNs) offer a simplified model of protein structure, dynamics, and communication pathways.
• Defining optimal criteria for constructing PSNs is crucial to include relevant contacts and exclude noise.

Purpose of the Study:

• To systematically evaluate residue distance cutoffs and contact probability thresholds for constructing informative protein structure networks.
• To identify objective criteria for PSN construction that minimize redundancy and maximize relevant information.

Main Methods:

• Constructed PSNs using atomistic molecular dynamics simulations across diverse proteins.
• Systematically varied residue distance cutoff lengths and contact formation probability thresholds.
• Quantified mutual information within resulting PSNs to assess network redundancy.

Main Results:

• A minimum in mutual information was consistently observed at a 5 Å residue distance cutoff, independent of the contact probability threshold.
• This finding held true across all distinct proteins analyzed in the study.
• The minimum mutual information indicates minimal redundancy in the PSN representation.

Conclusions:

• A 5 Å cutoff provides an objective and optimal criterion for constructing protein structure networks.
• This objective criterion supports the commonly used heuristic selection of 5 Å for PSN analysis.
• The findings enhance the reliability of PSNs for studying protein structure, dynamics, and internal communication.