Machine learning-assisted detection of single-point mutations via DNA-templated gold nanoparticle growth

Area of Science:

• Biotechnology
• Nanotechnology
• Molecular Diagnostics

Background:

• Accurate detection of single point mutations, like those in PIK3CA, is crucial for precision diagnostics but technically demanding.
• Existing methods often face challenges in sensitivity and specificity due to subtle genetic variations.

Purpose of the Study:

• To develop a machine learning-assisted colorimetric biosensor for sensitive and specific detection of single point mutations.
• To utilize DNA-templated gold nanoparticle (AuNP) growth for visual mutation identification.
• To integrate the biosensor with a smartphone-based AI platform for real-time analysis.

Main Methods:

• Employed a hairpin DNA probe that triggers AuNP growth and aggregation upon hybridization with mutant PIK3CA sequences.
• Utilized molecular dynamics simulations to understand AuNP-DNA interactions.
• Optimized critical parameters (DNA probe concentration, gold ion concentration, pH, temperature) using Plackett-Burman and Box-Behnken designs.
• Integrated the biosensor with a smartphone and a random forest regression model for AI-driven image analysis.

Main Results:

• Achieved a sensitive detection range of 10-1000 nM with a low detection limit of 8.13 nM for PIK3CA mutations.
• Demonstrated visible color changes (red to purple) correlating with mutation presence.
• The AI platform provided high recovery rates and low prediction errors for mutation detection.
• Molecular dynamics simulations revealed size-dependent AuNP binding to DNA structures.

Conclusions:

• The developed biosensor offers a scalable, cost-effective platform for single point mutation detection.
• Integration with AI enables portable, real-time analysis, enhancing potential for personalized diagnostics.
• The system shows promise for point-of-care testing, especially in resource-limited settings.