Deep Learning Enables Identification of Antimicrobial Peptides Through Mechanochromic Fingerprints

Area of Science:

• Biotechnology
• Materials Science
• Computational Biology

Background:

• Antimicrobial peptides (AMPs) show promise as alternatives to conventional antibiotics.
• Current methods for classifying AMP functions are slow and labor-intensive due to their diverse mechanisms of action.

Purpose of the Study:

• To develop a rapid, scalable, and accurate strategy for identifying and classifying antimicrobial peptides.
• To leverage polydiacetylene (PDA) mechanochromism, hyperspectral imaging, and deep learning for AMP screening.

Main Methods:

• Integration of low-cost, self-assembled polydiacetylene (PDA) sensors with hyperspectral imaging.
• Utilizing deep learning, specifically convolutional neural networks (CNNs), trained on full spectral datasets.
• Analyzing mechanochromic spectral fingerprints generated by AMP-PDA interactions.

Main Results:

• The developed strategy accurately distinguished seven AMPs at two concentrations with 96.79% accuracy.
• CNNs trained on full spectral data outperformed conventional two-wavelength colorimetric methods.
• Mechanochromic polymers were shown to encode significantly more chemical information than previously understood.

Conclusions:

• PDA mechanochromism, combined with high-throughput spectral imaging and deep learning, provides a powerful platform for rapid AMP screening.
• This approach establishes a foundation for scalable, information-dense biosensing technologies.
• The findings highlight the potential of mechanochromic polymers for detailed molecular interaction analysis.