Machine-Learning Single-Stranded DNA Nanoparticles for Bacterial Analysis

Abstract

A two-dimensional nanoparticle–single-stranded DNA (ssDNA) array has been assembled for the detection of bacterial species using machine-learning (ML) algorithms. Out of 60 unknowns prepared from bacterial lysates, 54 unknowns were predicted correctly. Furthermore, the nanosensor array, supported by ML algorithms, was able to distinguish wild-type Escherichia coli from its mutant by a single gene difference. In addition, the nanosensor array was able to distinguish untreated wild-type E. coli from those treated with antimicrobial drugs. This work demonstrates the potential of nanoparticle–ssDNA arrays and ML algorithms for the discrimination and identification of complex biological matrixes.

Machine Learning Enabled Nanosensor Array for Monitoring Citrus Juice Adulteration

“…The noncovalent interactions of F-ssDNA and 2D-np can be disrupted by a competing molecule or mixture of molecules resulting in the recovery of quenched fluorescence. The recovering fluorescence is highly dependent on the 2D-np type, F-ssDNA formulation, and the target type/mixture as demonstrated in our earlier studies. − Such diversity has allowed us to assemble biomarker-free nanosensor arrays against complex biological matrices including nucleic acids in urine, proteins in serum, and antibiotics in cellular lysates. − …”

“…The 2D-nps used in this study are nanographene oxide (nGO), molybdenum­(IV) disulfide (MoS 2 ), and tungsten­(IV) disulfide (WS 2 ), which are easy-to-synthesize and biologically stable nanomaterials that can be prepared in bulk scale. ,− The physiochemical properties of 2D-nps offer absorption of F-ssDNAs on their 2D surface through noncovalent forces while quenching the fluorescence of F-ssDNA. This property has been highly instrumental in the development of optical nanosensors against molecular targets or biological mixtures. − , …”

“…Citrus fruit juices being rich in various metabolites are an optimal target for our nanosensor array-based detection (Scheme a–c). − The citrus juice metabolites such as antioxidants, proteins, vitamins, and polyphenols will interact with both or either the 2D-np through noncovalent interactions. , As such, the F-ssDNA will then be displaced from the 2D-np surface through competition with the metabolite content of each juice type. − , The fraction of replaced F-ssDNA depends on the metabolite profile of each citrus juice type generating a fluorescence recovery pattern in response to various citrus juices or adulterants. Therefore, a different response is observed with each juice type.…”

“…This property has been highly instrumental in the development of optical nanosensors against molecular targets or biological mixtures. [34][35][36][37]42 The noncovalent interactions of F-ssDNA and 2D-np can be disrupted by a competing molecule or mixture of molecules resulting in the recovery of quenched fluorescence. The recovering fluorescence is highly dependent on the 2D-np type, F-ssDNA formulation, and the target type/mixture as demonstrated in our earlier studies.…”

“…Previously, we developed a fluorogenic single-stranded DNA (ssDNA) nanosensor-based biomarker-free platform in an array format to discriminate various biosystems such as breast cancer cells, bacteria species, miRNA, proteins, and antibiotics. − We further advanced this platform by coupling machine learning capabilities through an algorithmically guided optical nanosensor selector (AGONS) approach for superior prediction accuracy . Here, we show that an AGONS-enabled nanosensor array could be formatted for the detection of citrus juice adulterations through artificial sweetening or juice-to-juice dilution.…”