The study of intrinsically disordered proteins has rapidly advanced since the identification of the role they play in neurodegenerative diseases. Molecular dynamics simulations of disordered proteins have become common, but analysis tools optimized for their study have lagged behind. Both fully and partially disordered proteins present similar challenges: a vast fold space and difficultly in distinguishing meaningful protein motion. We have implemented an analysis tool based on inter-structure distance. This tool, g_isd, quantifies the differences between protein conformations. Our analysis is able to identify local regions that are flexible or disordered in otherwise folded proteins by employing a universal parameter that we developed to describe disorder. This order parameter has been scaled to be comparable between all proteins regardless or size or sequence length. We present one of the only clustering algorithms truly optimized to study protein dynamics. This hierarchical spectral clustering applies empirically-derived data to estimate meaningful protein motion allows unsupervised molecular dynamics clustering in reduced dimensional space. We apply our approach to the disordered loop region of a cystine knot protein. Analysis describes the dynamics of this loop containing a targeted binding sequence for the cancer-associated integrin alphavbeta6 protein. A sequence of steps to dock the cystine knot protein to its target as a large ligand is characterized. Finally, we analyze the disorder of a synthetic polymer with the useful property of thermal contraction. Molecular dynamics studies with a customized force field explain that a small difference in a single bond leads to significant disorder. The efficiency of thermal contraction can be modulated by varying levels of disorder in the material.