Andrzej Kolinski Research Group

Coarse-grained protein modeling

Modeling Software & Servers

Biomolecules — dynamics & interactions


Predicting real-valued protein residue fluctuation using FlexPred


Methods in Molecular Biology, 1484:175-186, 2017


The conventional view of a protein structure as static provides only a limited picture.There is increasing evidence that protein dynamics are often vital to protein function including interaction with partners such as other proteins, nucleic acids, and small molecules. Considering flexibility is also important in applications such as computational protein docking and protein design. While residue flexibility is partially indicated by experimental measures such as the B‐factor from X‐ray crystallography and ensemble fluctuation from nuclear magnetic resonance (NMR) spectroscopy as well as computational molecular dynamics (MD) simulation, these techniques are resource‐intensive. In this chapter, we describe the web server and standalone version of FlexPred, which rapidly predicts absolute per‐residue fluctuation from a three‐dimensional protein structure. On a set of 592 non‐redundant structures, comparing the fluctuations predicted by FlexPred to the observed fluctuations in MD simulations showed an average correlation coefficient of 0.669 and an average root mean square error of 1.07 Å. FlexPred is available at