Root Mean Square Fluctuation (RMSF)

Root Mean Square Fluctuation (RMSF) quantifies the time-averaged deviation of an individual atom (or group of atoms) from its average position. In molecular dynamics, it is primarily used to identify flexible and rigid regions of a macromolecule.

$$\text{RMSF}_i = \sqrt{\frac{1}{T} \sum_{t=1}^T \| \mathbf{r}_i(t) - \langle \mathbf{r}_i \rangle \|^2 }$$

Use Cases

• Flexibility Mapping: Identifying highly mobile loop regions versus rigid core secondary structures.
• B-factor Comparison: Converting theoretical RMSF values into isotropic B-factors ($B_i = \frac{8\pi^2}{3} \text{RMSF}_i^2$) for direct comparison with crystallographic temperature factors.
• Mutational Impact: Comparing the fluctuation profiles of wild-type and mutant proteins to spot localized changes in flexibility.

Pitfalls

• Lack of Alignment: Failing to remove global translation and rotation via a least-squares fit will artificially inflate fluctuations.
• Incorrect Reference: RMSF must be calculated relative to the average structure over the trajectory, not the starting structure.
• Atom Selection: Averaging RMSF over entire residues including highly flexible side-chains can mask the backbone’s true rigidity.

Code Snippets

1. NumPy / SciPy (Manual)

import numpy as np

def calc_rmsf(coords):
    # coords shape: (frames, atoms, 3)
    # Assumes trajectory is already aligned
    mean_structure = coords.mean(axis=0)
    fluctuations = coords - mean_structure
    return np.sqrt((fluctuations ** 2).sum(axis=-1).mean(axis=0))

2. VMD (Tcl)

set sel [atomselect top "protein and alpha"]
set num_frames [molinfo top get numframes]

# Align trajectory to the first frame first to remove global motion
set ref [atomselect top "protein and alpha" frame 0]
for {set i 0} {$i < $num_frames} {incr i} {
    $sel frame $i
    $sel move [measure fit $sel $ref]
}

# Calculate RMSF
set rmsf [measure rmsf $sel]
foreach atom [$sel get index] val $rmsf {
    puts "Atom $atom RMSF: $val"
}

3. MDAnalysis (Python)

import MDAnalysis as mda
from MDAnalysis.analysis import rms

u = mda.Universe("topology.pdb", "trajectory.xtc")
calphas = u.select_atoms("protein and name CA")

# rms.RMSF automatically aligns to the mean structure if requested
R = rms.RMSF(calphas).run()

# R.results.rmsf contains the output array