Prediction of Secondary Structure Content of Proteins Using Raman Spectroscopy and Self-Organizing Maps

Area of Science:

• Biophysics
• Spectroscopy
• Computational Biology

Background:

• Protein structure is crucial for function, but higher-order structure analysis is challenging.
• Traditional methods like Infrared (IR) and Circular Dichroism (CD) have limitations for concentrated or aqueous samples.
• Raman spectroscopy offers advantages for analyzing highly concentrated biopharmaceutical samples in aqueous solutions.

Purpose of the Study:

• To develop and validate a novel method for predicting protein secondary structure (SS) content.
• To assess the performance of Self-Organizing Maps (SOM) combined with Raman spectroscopy for SS analysis.
• To compare the accuracy of the Raman-SOM method against traditional techniques like band-fitting and Partial Least Squares (PLS) regression.

Main Methods:

• Collected back-scatter Raman spectra of proteins in both solid and aqueous states.
• Trained a Self-Organizing Maps (SOM) algorithm using spectral data with known SS content.
• Compared SOM predictions with results from Partial Least Squares (PLS) regression, band-fitting, and literature X-ray data.

Main Results:

• The Raman-SOM algorithm demonstrated high accuracy in predicting protein SS content.
• Prediction errors for the Raman-SOM method were comparable to PLS regression.
• The Raman-SOM method significantly outperformed traditional band-fitting techniques in accuracy.

Conclusions:

• Raman spectroscopy coupled with SOM is a viable and accurate alternative for protein secondary structure analysis.
• This method overcomes limitations of traditional techniques, especially for concentrated and aqueous biopharmaceutical samples.
• The developed Raman-SOM approach provides a robust tool for biopharmaceutical characterization and quality control.