Modeling amide-I vibrations of alanine dipeptide in solution by using neural network protocol

Area of Science:

• Computational Chemistry
• Spectroscopy
• Biophysics

Background:

• Infrared (IR) spectroscopy is vital for analyzing polypeptide structure and function.
• Ab initio calculations for interpreting IR spectra are computationally expensive.

Purpose of the Study:

• To develop a computationally efficient neural network (NN) protocol for evaluating polypeptide amide-I spectra.
• To reduce the cost associated with theoretical interpretation of IR spectra.

Main Methods:

• Density Functional Theory (DFT) calculations were used to obtain structural parameters and amide-I frequencies for alanine dipeptide (ALAD) conformers.
• A neural network (NN) model was trained using DFT data across various micro-environments.
• The NN protocol was applied to predict ALAD amide-I frequencies in different solvation conditions.

Main Results:

• The developed NN protocol accurately predicted amide-I frequencies for ALAD in various environments.
• The NN approach significantly reduced computational costs compared to traditional electronic structure calculations.
• The model demonstrated satisfactory performance in predicting spectra under different solvation conditions.

Conclusions:

• The NN modeling protocol offers a cost-effective alternative for analyzing polypeptide secondary structures via IR spectroscopy.
• This approach facilitates the study of dynamic secondary structures and biological functions using backbone vibrational probes.
• The findings enable efficient deciphering of polypeptide behavior through vibrational spectroscopy.