CheSPI: chemical shift secondary structure population inference

NMR chemical shifts (CSs) are delicate reporters of local protein structure, and recent advances in random coil CS (RCCS) prediction and interpretation now offer the compelling prospect of inferring small populations of structure from small deviations from RCCSs. Here, we present CheSPI, a simple and efficient method that provides unbiased and sensitive aggregate measures of local structure and disorder. It is demonstrated that CheSPI can predict even very small amounts of residual structure and robustly delineate subtle differences into four structural classes for intrinsically disordered proteins. For structured regions and proteins, CheSPI provides predictions for up to eight structural classes, which coincide with the well-known DSSP classification. The program is freely available, and can either be invoked from URL as a web implementation, or run locally from command line as a python program. CheSPI generates comprehensive numeric and graphical output for intuitive annotation and visualization of protein structures. A number of examples are provided.

Automated summary

CheSPI is a simple and efficient method that can accurately predict local structure and disorder, even in very small amounts of residual structure. It provides predictions for up to eight structural classes for structured regions and proteins, and generates intuitive numeric and graphical output.