Intrinsically disordered proteins (IDPs) are distinguished by their lack of a stable,
well-organized three-dimensional conformation [1].
Conventional structure prediction algorithms collapse dynamic conformational ensembles
into static, inappropriately compacted, or inaccurate structural representations [2].
We present a computational pipeline engineered to facilitate efficient generation
of conformationally heterogeneous and biophysically valid initial ensembles appropriate
for subsequent large-scale molecular dynamics (MD) simulations.
Our approach implements conventional self-avoiding walk Cartesian constraints through
systematic sampling of biophysically permissible Ramachandran (ϕ⁄ψ) dihedral angle
distributions utilizing idealized peptide bond geometry.
Steric incompatibilities are methodically resolved during preliminary workflow stages
via an in-memory vacuum minimization protocol utilizing OpenMM [3], preceding computationally
intensive explicit solvent treatment procedures.
To optimize computational efficiency, systems underwent solvation and simulation
employing Hydrogen Mass Repartitioning (HMR) [4] within GROMACS [5], enabling an accelerated 4-fs
integration time-step while maintaining rigorous thermal and potential energy
equilibration at 300 K.
Utilizing this workflow, a 100-model ensemble of the 95-amino acid core domain of
the TAU8 isoform (TAU-F) was successfully constructed.
Reference:
1. Harake, Sami N. Al, et al. “Beyond the Structure-Function Paradigm: A Comprehensive Review of Intrinsically Disordered Proteins.” Biochemistry and Biophysics Reports, vol. 47, July 2026, p. 102706. PubMed Central, https://doi.org/10.1016/j.bbrep.2026.102706.
2. . Ruff, Kiersten M., and Rohit V. Pappu. “AlphaFold and Implications for Intrinsically Disordered Proteins.” Journal of Molecular Biology, vol. 433, no. 20, Oct. 2021, p. 167208. PubMed, https://doi.org/10.1016/j.jmb.2021.167208.
3. Eastman, P., et al. "OpenMM 8: Molecular Dynamics Simulation with Machine Learning Potentials." J. Phys. Chem. B 128(1), 109-116 (2023).
4. Hopkins, C. W., Le Grand, S., Walker, R. C., & Roitberg, A. E. (2015). Long-Time-Step Molecular Dynamics through Hydrogen Mass Repartitioning. Journal of Chemical Theory and Computation, 11(4), 1864–1874.
5. Abraham, M. J., Murtola, T., Schulz, R., Páll, S., Smith, J. C., Hess, B., & Lindahl, E. (2015). GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX, 1–2, 19–25.
Notes:
-
The Energy, Pressure, and Temperature parameters do not commence from zero.
This results from the incorporation of two preparatory computational steps
within the refined workflow that precede the relaxation phase—the most computationally
intensive and time demanding stage of the entire process. These two implemented steps
comprise OpenMM Vacuum Relaxation and Temperature Warming protocols. These modifications
effectively mitigate atomic collision phenomena, specifically the abrupt escalation of
intermolecular forces to near-infinite magnitudes. Consequently, the accompanying Energy
and Temperature graphical representations demonstrate solely the oscillations of these
parameters throughout the relaxation phase, during which the Root Mean Square Deviation
(RMSD) analysis already indicates protein structural stabilization.
-
The pressure dynamics present a distinct consideration. Pressure stabilization procedures
were not implemented in this pipeline, nor are they requisite for our objectives. The
interatomic bonds exhibit sufficient strength that pressure fluctuations would exert
negligible influence on structural folding. The appropriate methodological sequence involves
initial molecular docking with small molecule ligands or design of attached peptide. Subsequently,
if warranted, structural relaxation of the complete system could be undertaken. However,
such relaxation is not necessary for our current purposes. Our investigation is specifically
directed toward obstructing the particular contact interface rather than identifying the final
equilibrium configuration of the compound.
The pressure dynamics present a distinct consideration. Pressure stabilization procedures
were not implemented in this pipeline, nor are they requisite for our objectives.
From a practical perspectives, the structural relaxation of the complete system should be undertaken.
The Radius Gyration (R
g) plot shows a wide structural distribution typical of an intrinsically
disordered protein (IDP) ensemble, ranging from highly compact to highly extended states.
While 100 conformations are a small snapshot compared to the thousands typically needed for
a fully converged IDP ensemble, this initial batch gives a clear look at the structural
landscape.
This particular plot Shows:
-
Wide Structural Variance: The individual folds span an Rg from roughly 1.37 nm to 4.98 nm.
For an IDP, this vast range is expected and normal. It indicates that the generation method is properly
sampling different states rather than getting stuck in a single rigid conformation.
-
The Two Extremes:
-
Folds near 1.37 nm represent highly collapsed, compact globular-like states.
-
Folds near 4.98 nm represent highly extended, random-coil-like states.
-
Ensemble Central Tendency: The red dashed line shows your mean Rg sits at 2.30 nm.
The light pink band shows where the majority of your structures fall (+/- 1 Standard Deviation).