Allosteric signaling in biomolecules is a key mechanism for a myriad of cellular processes. We present a general yet compact model for protein allostery at atomic detail to quantitatively explain and predict structural-dynamics properties of allosteric signal propagation. The master equation-based approach for allostery by population shift (MAPS) is introduced that derives the time scales, amplitudes, and pathways of signal transmission in peptides and proteins from dihedral angle dynamics observed in extended molecular dynamics simulations. The MAPS approach is first applied to the alanine-pentapeptide, and the results are tested against an explicit simulation in the presence of local conformational constraints, confirming the validity and accuracy of the model. We then apply the approach to a larger Markovian system based on a millisecond all-atom protein molecular dynamics trajectory of BPTI (Shaw et al. Science 2010, 330, 341-346). We use MAPS to illustrate in silico the propagation of a local perturbation over medium- to long-range distances across a disulfide bridge linking loops L1 and L2, which constitute the binding interface of BPTI.