Model based variable selection as a tool to highlight biological differences in Raman spectra of cells

Area of Science:

• Biophysics
• Biochemistry
• Spectroscopy

Background:

• In vitro Raman spectroscopy offers non-invasive characterization of cells and tissues.
• Cell and tissue spectra are complex, with overlapping signals from lipids, proteins, carbohydrates, and nucleic acids, hindering analysis.
• Discerning molecular components responsible for spectral changes is challenging.

Purpose of the Study:

• To develop a novel method for highlighting significant biochemical changes between sample groups in Raman spectroscopy.
• To improve the discrimination of cellular states based on biomolecular differences.

Main Methods:

• Application of Partial Least Squares - Discriminant Analysis (PLS-DA) Variable Importance Projection (VIP) scores as heat maps.
• Combination of VIP scores with group difference spectra to identify significant Raman spectral changes.
• Analysis of single HeLa cells in various states: live, fixed, fixed and dehydrated, and fixed, dehydrated, and rehydrated.

Main Results:

• The novel technique successfully highlighted significant biochemical changes between different cellular states.
• Specific molecular changes induced by fixation, dehydration, and rehydration were identified.
• The method effectively visualized differences in Raman band shapes and positions.

Conclusions:

• The developed method provides a powerful approach to interpret complex Raman spectra from biological samples.
• This technique enhances the ability to identify and understand molecular alterations in cells under different conditions.
• It offers improved insights into cellular biochemical processes and biomolecular differences.