Structural ensemble reconstruction for Intrinsically disordered proteins via a physics-based coarse-grained model guided by NMR parameters

Kierownik projektu: Adam Liwo


  • Yi He

Uniwersytet Gdański

Wydział Chemii


Data otwarcia: 2021-08-13

Streszczenie projektu

Flexible and intrinsically disordered proteins (IDPs) play critical roles in signal propagation in biological systems. IDPs are recognized by their structural heterogeneity in conformational space. While current experimental techniques provide ensemble-averaged properties of IDPs, they are unable to capture dynamic properties involving a large conformational space. Molecular dynamics (MD) simulations are an alternative method to investigate IDPs by providing conformational trajectories at atomic- or residue-level resolution. Given the heterogeneity of IDP ensembles, extensive simulations are required to cover multiple local minima on the energy landscape. Therefore, simulating IDPs is much more computationally demanding than simulating folded proteins. Recently, Shaw and co-workers have demonstrated that all-atom molecular dynamics simulations are a reliable approach to sampling heterogeneous structures of disordered proteins. While these results are promising, this requires extensive computational resources. Physics-based coarse-grained force fields, such as the United-RESidue (UNRES) force field, can be a good alternative to reduce the computational cost and achieve better sampling of the conformational space of IDPs. In the UNRES model, ⍺-carbons define the geometry of polypeptide chains. The united peptide groups and the united side chains act as interaction sites. The energy function of the UNRES model is defined as a potential of the mean force of proteins in water, where secondary degrees of freedom have been averaged out. We have demonstrated that the UNRES force field can successfully predict the structures of folded proteins and has achieved excellent performance in Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiments. If experimental data are used to optimize the UNRES results of IDPs, this may improve the accuracy of sampled conformations of unstructured proteins.
Because chemical shifts (CS) from NMR experiments can be measured under a wide range of conditions with great accuracy, CS have become the most accessible parameter for structural characterization of flexible proteins. Furthermore, the incorporation of experimental data in computational modeling can improve the accuracy of computational predictions and ultimately fill the gap of missing information. Recently, we have developed a single-residue based chemical shift-protein structure database and a complimentary program called Glutton. Glutton is about 1000 times faster than an all-atom sampling of the conformation space. Moreover, Glutton can identify multiple structural preferences corresponding to only one set of chemical shifts for residues which fluctuated between multiple states rather than only one structure. With the torsional angle restrictions obtained from Glutton, we will modify the UNRES force field by including γ angle distributions to achieve a more complete search of the conformational space using multiplexed replica exchange molecular dynamics (MREMD). MREMD has been proven to better sample the conformational space, as demonstrated in previous works. Furthermore, we will include an additional energy term to better reflect the sampling of unstructured proteins. A fine-grained cluster analysis based on free energies will be used to obtain a structural ensemble of 10,000 structures. The developed chemical shift-UNRES program will be validated using ASC protein experimental data provided by Dr. Eva De Alba at the University of California Merced. The chemical shift data will be used to guide the simulations, and the Residual Dipolar Coupling (RDC) and Nuclear Overhauser effect (NOE) data will be used to validate simulation results independently.
Our aim is to optimize the UNRES force field so that it can effectively and accurately sample the conformational space of disordered proteins. In addition, we will develop a strategy to guide UNRES simulations with dihedral angle distributions extracted from NMR chemical shift data via Glutton. With the above goals, we would like to request access to Tryton supercomputer and related storage system. This access will allow us to benchmark and optimize the UNRES force field and, ultimately, test the final strategy of NMR chemical shift-guided UNRES simulations.
The TASK supercomputer has the developer/latest version of UNRES, which is critical for the completion of the project. The computational power provided for the Tryton supercomputer is needed for optimization and extensive evaluation of the optimized UNRES force fields with real-world applications.

← Powrót do spisu projektów


Nasi konsultanci służą pomocą przyszłym i początkującym użytkownikom specjalistycznego oprogramowania zainstalowanego na Komputerach Dużej Mocy w Centrum Informatycznym TASK.

Kontakt w sprawach Komputerów Dużej Mocy, oprogramowania/licencji, grantów obliczeniowych, sprawozdań:

Administratorzy odpowiadają na maile w dni robocze w godzinach 8:00 – 15:00.