Are there non-trivial dynamic cross-correlations in proteins?

Area of Science:

• Molecular dynamics
• Computational biology
• Structural biology

Background:

• Analyzing internal molecular motion necessitates removing overall motion via least-squares fitting.
• The selection of atoms for fitting impacts the observed internal motion, as seen with BPTI.
• Molecular dynamics simulations are crucial for understanding protein flexibility.

Purpose of the Study:

• To investigate the impact of atom selection in fitting procedures on essential dynamics analysis.
• To introduce and evaluate a novel method, principal component analysis in distance space, for analyzing molecular motion.
• To determine if distance space analysis preserves dynamic cross-correlations without overall motion bias.

Main Methods:

• Performed essential dynamics (ED) analyses on a 1 ns molecular dynamics trajectory of a single-stranded DNA-binding protein.
• Compared ED results using all alpha-carbon atoms versus only the least mobile atoms for trajectory fitting.
• Developed and applied principal component analysis (PCA) in distance space to analyze dynamic cross-correlations.

Main Results:

• Fitting with least mobile atoms significantly reduced long-range atomic covariances compared to using all alpha-carbons.
• Despite reduced covariances, the overlap between essential spaces remained high.
• Principal component analysis in distance space successfully preserved non-trivial dynamic cross-correlations.

Conclusions:

• The choice of fitting atoms influences the analysis of internal molecular motion, particularly long-range covariances.
• Principal component analysis in distance space offers a viable method to analyze molecular dynamics without bias from overall motion.
• Distance space analysis preserves essential dynamic information, though cross-correlations may be smaller than in standard ED.