· Delarue web servers
  · Introduction
· Coarse-Graining
  · Protein CG
· Non-constant Kij
  · Hessian Proj. CG
  · Decimation CG
· New NMA Models
  · Go-like Energy NMA
  · Delaunay + classic NMA
· References
· Links
· Job queue status
· Distances Bridges

Welcome to DD-NMA website (beta-version)

This site provides tools for calculating Normal Modes using either Full-atomic or Coarse-Grained Models of Proteins and Nucleic Acids using a variant of the Elastic Network, with no adjustable parameter.

It allows to calculate Normal Modes for very large macromolecular assemblies as large as complete viral capsids (Koehl and Delarue, 2019). It is the collaborative work of Fred Poitevin, Rafael Navaza, Marc Delarue and Patrice Koehl.

On one hand it permits to define the Network based on Delauney representation of proteins, and it does not rely on any cutoff to define the neighbours of each node (Ref 1).

On the other hand, it allows several ways to coarse-grain the molecule, from the simplest method (Refs 2-3) to the more sophisticated and rigourous Decimation method, described here.
Hence the name DD-NMA.

The code for Normal Modes calculation itself has been rewritten to accommodate very large structures, comprising up to than 300,000 atoms.
A possible application handling very large macromolecular complexes such as entire viruses is described in Frontiers in Molecular Biosciences, 2017.

One novelty is a more systematic way to achieve coarse-graining.
-One trivial way is to select atoms (beads) based on their names in each amino-acid. Indeed we provide services that allows to select a few representative points per amino-acid (2), or nucleotide (3-5).
-A more rigourous "projection" of the Hessian of an all-atoms elastic model onto a CA-only network whose force constants are re-calculated has been implemented. It proceeds along lines described by Na and Song (6) and others.
-Additionnally, we implemented a method that "prunes" the Network beyond the CA-only bead model by the so-called Decimation method (see Illustration here), using a rigourous algorithm designed to keep the low-frequency modes of the model (Koehl et al., 2017).

Another novelty, compared to other web-sites-based methods based on Elastic Networks, is the use of Delaunay tessalation to define linked neighbours in the Network Model (1), instead of relying on a rather arbitrary radial cutoff.

Finally, the last major novelty is the implementation of the calculation of Normal Modes using a Hessian derived from a more complete Go-like energy, that includes not only a term on distances, but also three additional elastic terms based on three-body and four-body (dihedral) angles (7).

In this way, we depart from the classical Elastic Network defined originally by Monique Tirion, while keeping most of its salient features (8).

Key References can be found here:
1. Patrice Koehl's method and code for NMA and Delaunay.
2. 2-3 beads per amino-acid: simplified representation of proteins.
3. 6-7 beads per nucleotide: simplified representation of nucleic acids.
4. Elastic Models and RNA structures.
5. Comparison of Elastic Models for RNA with MD and experiments.
6. Rigourous projection method onto CA-only model.
7. Improved Elastic Network with a Go-like Energy.
8. Universality of vibrational spectra of Proteins.

This is a Beta version web site - still under development....

  Marc Delarue http://lorentz.dynstr.pasteur.fr
Page last modified 21:14 October 29, 2020.