Dissacharide Mapping
DiMap was created as a practical alternative to metadynamics for exploring disaccharide conformations around the same glycosidic collective variables (φ/ψ). While metadynamics can provide free-energy surfaces, it typically requires long simulations, bias-parameter tuning, and convergence checks. DiMap focuses on a simpler and reproducible strategy to quickly map the relevant torsional space.
DiMap (energy-based torsional map)
|
Metadynamics (free-energy surface)
|
Note: DiMap reports a minimized potential-energy landscape (CHARMM36 + implicit solvent), whereas metadynamics yields a free-energy surface (ΔG). Even so, the location of the main low-energy basins/minima can be compared qualitatively across both maps.
DiMap generates an energy-based torsional map of a disaccharide by:
- scanning the φ/ψ space with a user-defined rotation step (e.g., 2.5°, 5°, 10°),
- rotating coordinates to generate candidate conformations,
- minimizing each conformation and evaluating its energy using NAMD (CHARMM36 + implicit solvent),
- reporting the resulting landscape and highlighting local/global minima.
These minima can then be used as starting conformations to build larger carbohydrate units (e.g., repetitive units with substituents) and to launch subsequent MD simulations.
The table below summarizes the wall-clock time required to generate φ/ψ maps for a set of 8 disaccharides (D1–D8; ~44–51 atoms), using three map resolutions (φ/ψ step size). Values are means across the benchmark set:
| Map resolution (φ/ψ step) | DiMap (mean wall time) | Metadynamics (mean wall time) | Mean speedup |
|---|---|---|---|
| 2.5° | 1h 49m (6,585 s) | 10h 46m (38,774 s) | 5.9× |
| 5° | 27m 2s (1,622 s) | 9h 44m (35,068 s) | 21.6× |
| 10° | 6m 48s (408 s) | 9h 48m (35,298 s) | 86.6× |
Time ranges across the benchmark set:
- 2.5°: DiMap 1h 47m–1h 59m, metadynamics 10h 17m–11h 33m.
- 5°: DiMap 26m 8s–29m 52s, metadynamics 9h 25m–10h 26m.
- 10°: DiMap 6m 35s–7m 32s, metadynamics 9h 27m–10h 29m.