Machine Learning Enhances Single-Molecule Sensing with Nanoparticle Conjugates

Researchers from the University of New South Wales and Australian Centre for Nanomedicine have developed a new technique for analyzing the plasmonic resonance shift of nanoparticle conjugates with the help of machine learning (1). The team was led by J. Justin Gooding, and their work was published in the journal Analytical Chemistry (1).

Plasmonic nanoparticles are commonly used in single-molecule sensing, where they are arranged in a dimer format. When a target molecule interacts with hairpin DNA, it causes a shift in the interparticle distance and results in a localized surface plasmon resonance shift. This shift can be detected using spectroscopy, but it requires the measurement of thousands of nanoparticle dimers, which is time-consuming and incompatible with point-of-care devices.

To overcome this challenge, the researchers used dark-field imaging of the dimer structures and analyzed the plasmonic resonance shift using machine learning (1). By digitally separating dimers from other nanoconjugate types, the team reduced false signals caused by non-specifically bound clusters of nanoparticles (1).

The team observed that the variation in image intensity had a discernible impact on the color analysis of the nanoconjugate constructs and thus the accuracy of the digital separation (1). To address this issue, the team compared different color spaces, including RGB, HSV, and LAB, to train a classifier algorithm (1). The LAB-based learning classifier demonstrated the highest accuracy for digitally separating nanoparticles (1).

Using the LAB-based learning classifier, the team monitored the plasmonic color shift of nanoparticle conjugates after interacting with a synthetic RNA target (1). The platform showed a highly accurate yes-or-no response with a true positive rate of 88% and a true negative rate of 100% (1). The sensor response of tested single-stranded RNA samples was well above control responses for target concentrations in the range of 10 aM–1 pM (1).

In summary, this new technique provides a promising approach for enhancing single-molecule sensing with nanoparticle conjugates. The ability to achieve statistical relevance with high throughput in minutes makes this technique suitable for point-of-care devices, which could revolutionize the field of diagnostic testing.

Reference

(1) Bennett, D.; Chen, X.; Walker, G. J.; Stelzer-Braid, S.; Rawlinson, W. D.; Hibbert, D. B.; Tilley, R. D.; Gooding, J. J. Machine Learning Color Feature Analysis of a High Throughput Nanoparticle Conjugate Sensing Assay. Anal. Chem. 2023, ASAP. DOI: 10.1021/acs.analchem.2c05292