Detecting intermediate structures in protein conformational pathways

In this work we present a topology and geometry-based analysis of the conformational space of proteins. Our goal is detecting clusters of highly populated, low-energy structures that may correspond to interesting intermediate states which are hard to characterize experimentally. We used coarse-grained conformational sampling and a topological software package, JavaPlex, to characterize distinct clusters of possible intermediate structures. This information can help us determine the structural and hopefully functional properties of these intermediates. We applied our method to three medium to large proteins that undergo large-scale conformational changes. Our results are consistent with available experimental data and demonstrate our ability to efficiently cluster the conformations while filtering out noise and outliers. Additionally, algebraic topology can be used to obtain information about the robust topological properties of the conformational landscape of proteins.