When compared to NMA, ENM does not require energy minimization of the input protein structure.Thus, is it advisable to perform EM before ENM analysis especially in case of homology modeled structures or would NMA be more reliable?
It is advised to perform energy minimization for any Normal Mode Analysis you might want using analytical gradients, whether it's using Molecular Mechanics force fields, quantum mechanical potentials or the elastic network model. Normal Modes only make sense in the context of the harmonic oscillator approximation, so you need to have a structure that is at its minimum in order to try anything. The Elastic Networks Model assumes that your structure represents the equilibrium point in the oscillations, and thus a minimum energy structure. So yes, it is highly advisable that you perform an energy minimization before ENM analysis.
The complete normal mode description (which you're referring to as NMA) is always more accurate than the normal mode decomposition done in the elastic model approximation, however it has been shown that for low frequency normal modes (those that represents oscillations of the protein as a whole, as opposed to those describing more local motion) can be well described by the elastic model network approximation, and also the ENM is quite fast. So it really depends on what you want your normal modes for. If you want normal modes to describe general conformational changes, then I'd stick to ENM, however if certain hinge and/or active site SPECIFIC motion is of interest, I'd think about doing an NMA.
It is advised to perform energy minimization for any Normal Mode Analysis you might want using analytical gradients, whether it's using Molecular Mechanics force fields, quantum mechanical potentials or the elastic network model. Normal Modes only make sense in the context of the harmonic oscillator approximation, so you need to have a structure that is at its minimum in order to try anything. The Elastic Networks Model assumes that your structure represents the equilibrium point in the oscillations, and thus a minimum energy structure. So yes, it is highly advisable that you perform an energy minimization before ENM analysis.
The complete normal mode description (which you're referring to as NMA) is always more accurate than the normal mode decomposition done in the elastic model approximation, however it has been shown that for low frequency normal modes (those that represents oscillations of the protein as a whole, as opposed to those describing more local motion) can be well described by the elastic model network approximation, and also the ENM is quite fast. So it really depends on what you want your normal modes for. If you want normal modes to describe general conformational changes, then I'd stick to ENM, however if certain hinge and/or active site SPECIFIC motion is of interest, I'd think about doing an NMA.
NMA obtains the normal modes by calculating the second derivative of the potential energy evaluated at a stable/minimized structure. In ENM the protein is modeled as a spring network in which all springs have the same force constant (again a stable/minimized structure should be used). Both methods obtain the normal modes through the same method:
1. Generate a 3N by 3N (N=number of atoms in protein - typically only alpha carbons are used) matrix by evaluating the change in the potential energy as each atom is moved a little bit in each direction (the second derivative of the potential energy).
2. Diagonalize the (mass-weighted) matrix to obtain eigenvectors (the modes) and eigenvalues (the frequencies).
The key difference is in the assumptions in building the models: NMA in principle retains all the information from a detailed, typical MD potential function; ENM uses a simplified potential which ignores the details of the potential and neglects long-range interactions. Of course, the obvious advantage of ENM is that it is much faster; it tends to perform best (i.e. matching the results from simulations) for globular proteins and for buried residues (because the spring network is a more reliable approximation for atoms/residues that are in compact environments).
I attach a presentation I gave a few years ago comparing the basic differences among the various methods for calculating normal modes.
i hope this is helpful.
- Badges
- Science topic
- Similar topics
- Chemistry
- Theoretical Chemistry
- Computational Chemistry
- Molecular Dynamics