Mapping the intramolecular signal propagation pathways in protein using Bayesian change point analysis of atomic motions.

We propose to use change points of atomic positions in the molecular dynamics trajectory as indicators of the propagating signals in protein. We designate these changes as signals because they can propagate within the molecule in the form of "perturbation wave", transmit energy or information between different parts of protein, and serve as allosteric signals. We found that change points can distinguish between thermal fluctuations of atoms (noise) and signals in a protein despite the differences in the motility of amino acid residues. Clustering of the spatially close residues that were experiencing change points close in time, allowed us to map pathways of signal propagation in a protein at the atomic level of resolution. We propose a potential mechanism for the origin of the signal and its propagation that relies on the autonomic coherence resonance in atomic fluctuations. According to this mechanism, random synchronization of fluctuations of neighboring atoms results in a resonance, which increases amplitude of vibration of these atoms. This increase can be transmitted to the atoms colliding with the resonant atoms, leading to the propagating signal. The wavelet-based coherence analysis of the inter-atomic distances between carbon-alpha atoms and surrounding atoms for the residue pairs that belong to the same communication pathway allowed us to find time periods with temporarily locked phases, confirming the occurrence of conditions for resonance. Analysis of the mapped pathways demonstrated that they form a network that connects different regions of the protein.