Machine Learning-Assisted DNA Origami Shape Sorting Using Fingerprinting Nanosensors and Feature Engineering.

Area of Science:

• Nanotechnology and Molecular Engineering
• Biomolecular Engineering
• Machine Learning Applications

Background:

• Reconfigurable DNA nanostructures are vital for applications like drug delivery and biosensing.
• Current methods for verifying DNA nanostructure folding, such as electron microscopy, are complex and costly.
• There is a need for accessible, low-cost techniques to confirm DNA nanostructure assembly and shape.

Purpose of the Study:

• To develop a novel, low-cost nanosensor array for distinguishing between different DNA origami shapes.
• To integrate machine learning for accurate classification of DNA nanostructures.
• To provide a generalizable platform for high-throughput, label-free shape sorting in DNA origami.

Main Methods:

• A nanosensor array was created using graphene oxide nanosheets complexed with fluorophore-labeled DNA probes.
• DNA nanostructures were introduced to the array, causing fluorescence recovery signatures due to probe displacement.
• Machine learning algorithms were employed to analyze these signatures for shape classification.

Main Results:

• The nanosensor array successfully differentiated between DNA origami triangle and nanotube shapes, as well as unfolded scaffold strands.
• Achieved a prediction accuracy of 94% in discriminating between the three DNA configurations.
• Demonstrated a unique fluorescence recovery signature for each DNA nanostructure type.

Conclusions:

• The developed fingerprinting nanosensor array with machine learning provides an effective method for DNA nanostructure shape sorting.
• This approach offers a valuable, high-throughput, and label-free alternative to conventional microscopy techniques.
• The platform is generalizable, opening new possibilities for characterizing DNA nanostructures.