Metadynamics and Raman Spectroscopy for Glycan Structure-Spectrum Mapping.

Area of Science:

• Biophysical Chemistry
• Computational Chemistry
• Spectroscopy

Background:

• Raman spectroscopy provides label-free molecular vibration data across biological systems.
• Interpreting Raman spectra for complex biomolecules like glycans is challenging due to heterogeneity.
• Existing methods struggle to link conformational ensembles to vibrational spectra.

Purpose of the Study:

• To introduce DynaSpec, a novel framework for ensemble-resolved structure-spectrum mapping.
• To enable mechanistic interpretation of complex Raman spectra by linking molecular conformations to vibrational data.
• To facilitate bidirectional inference between biomolecular structure and Raman spectra.

Main Methods:

• Integrating metadynamics in a machine-learned latent space with population-weighted DFT Raman calculations.
• Utilizing experimental spectroscopy alongside computational predictions for validation.
• Developing a framework for bidirectional structure-spectrum inference.

Main Results:

• DynaSpec achieved >85% classification accuracy for 13 N-glycans.
• Identified a conserved α1,6 open-curled switching motif in N-glycans.
• Enabled accurate discrimination of glycan positional isomers (>70%) and unmixing of mixtures (R² > 0.99).
• Demonstrated generalization to O-glycans and glycosaminoglycans, validated with hyaluronic acid.

Conclusions:

• DynaSpec establishes an ensemble-resolved framework for structure-informed spectroscopy.
• The framework accurately interprets congested Raman signatures and links them to molecular conformations.
• This approach advances the analysis of diverse biomolecular classes using vibrational spectroscopy.