Utilizing Deep Learning Algorithms for Signal Processing in Electrochemical Biosensors: From Data Augmentation to Detection and Quantification of Chemicals of Interest.

Nanomaterial-based aptasensors serve as useful instruments for detecting small biological entities. This work utilizes data gathered from three electrochemical aptamer-based sensors varying in receptors, analytes of interest, and lengths of signals. Our ultimate objective was the automatic detection and quantification of target analytes from a segment of the signal recorded by these sensors. Initially, we proposed a data augmentation method using conditional variational autoencoders to address data scarcity. Secondly, we employed recurrent-based networks for signal extrapolation, ensuring uniform signal lengths. In the third step, we developed seven deep learning classification models (GRU, unidirectional LSTM (ULSTM), bidirectional LSTM (BLSTM), ConvGRU, ConvULSTM, ConvBLSTM, and CNN) to identify and quantify specific analyte concentrations for six distinct classes, ranging from the absence of analyte to 10 µM. Finally, the second classification model was created to distinguish between abnormal and normal data segments, detect the presence or absence of analytes in the sample, and, if detected, identify the specific analyte and quantify its concentration. Evaluating the time series forecasting showed that the GRU-based network outperformed two other ULSTM and BLSTM networks. Regarding classification models, it turned out signal extrapolation was not effective in improving the classification performance. Comparing the role of the network architectures in classification performance, the result showed that hybrid networks, including both convolutional and recurrent layers and CNN networks, achieved 82% to 99% accuracy across all three datasets. Utilizing short-term Fourier transform (STFT) as the preprocessing technique improved the performance of all datasets with accuracies from 84% to 99%. These findings underscore the effectiveness of suitable data preprocessing methods in enhancing neural network performance, enabling automatic analyte identification and quantification from electrochemical aptasensor signals.

Anomaly Detection for Sensor Signals Utilizing Deep Learning Autoencoder-Based Neural Networks.

Anomaly detection is a significant task in sensors' signal processing since interpreting an abnormal signal can lead to making a high-risk decision in terms of sensors' applications. Deep learning algorithms are effective tools for anomaly detection due to their capability to address imbalanced datasets. In this study, we took a semi-supervised learning approach, utilizing normal data for training the deep learning neural networks, in order to address the diverse and unknown features of anomalies. We developed autoencoder-based prediction models to automatically detect anomalous data recorded by three electrochemical aptasensors, with variations in the signals' lengths for particular concentrations, analytes, and bioreceptors. Prediction models employed autoencoder networks and the kernel density estimation (KDE) method for finding the threshold to detect anomalies. Moreover, the autoencoder networks were vanilla, unidirectional long short-term memory (ULSTM), and bidirectional LSTM (BLSTM) autoencoders for the training stage of the prediction models. However, the decision-making was based on the result of these three networks and the integration of vanilla and LSTM networks' results. The accuracy as a performance metric of anomaly prediction models showed that the performance of vanilla and integrated models were comparable, while the LSTM-based autoencoder models showed the least accuracy. Considering the integrated model of ULSTM and vanilla autoencoder, the accuracy for the dataset with the lengthier signals was approximately 80%, while it was 65% and 40% for the other datasets. The lowest accuracy belonged to the dataset with the least normal data in its dataset. These results demonstrate that the proposed vanilla and integrated models can automatically detect abnormal data when there is sufficient normal data for training the models.

Predicting Analyte Concentrations from Electrochemical Aptasensor Signals Using LSTM Recurrent Networks.

Nanomaterial-based aptasensors are useful devices capable of detecting small biological species. Determining suitable signal processing methods can improve the identification and quantification of target analytes detected by the biosensor and consequently improve the biosensor's performance. In this work, we propose a data augmentation method to overcome the insufficient amount of available original data and long short-term memory (LSTM) to automatically predict the analyte concentration from part of a signal registered by three electrochemical aptasensors, with differences in bioreceptors, analytes, and the signals' lengths for specific concentrations. To find the optimal network, we altered the following variables: the LSTM layer structure (unidirectional LSTM (LSTM) and bidirectional LSTM (BLSTM)), optimizers (Adam, RMSPROP, SGDM), number of hidden units, and amount of augmented data. Then, the evaluation of the networks revealed that the highest original data accuracy increased from 50% to 92% by exploiting the data augmentation method. In addition, the SGDM optimizer showed a lower performance prediction than that of the ADAM and RMSPROP algorithms, and the number of hidden units was ineffective in improving the networks' performances. Moreover, the BLSTM nets showed more accurate predictions than those of the ULSTM nets on lengthier signals. These results demonstrate that this method can automatically detect the analyte concentration from the sensor signals.

Synthesis of encapsulated ZnO nanowires provide low impedance alternatives for microelectrodes.

Microelectrodes are commonly used in electrochemical analysis and biological sensing applications owing to their miniaturised dimensions. It is often desirable to improve the performance of microelectrodes by reducing their electrochemical impedance for increasing the signal-to-noise of the recorded signals. One successful route is to incorporate nanomaterials directly onto microelectrodes; however, it is essential that these fabrication routes are simple and repeatable. In this article, we demonstrate how to synthesise metal encapsulated ZnO nanowires (Cr/Au-ZnO NWs, Ti-ZnO NWs and Pt-ZnO NWs) to reduce the impedance of the microelectrodes. Electrochemical impedance modelling and characterisation of Cr/Au-ZnO NWs, Ti-ZnO NWs and Pt-ZnO NWs are carried out in conjunction with controls of planar Cr/Au and pristine ZnO NWs. It was found that the ZnO NW microelectrodes that were encapsulated with a 10 nm thin layer of Ti or Pt demonstrated the lowest electrochemical impedance of 400 ± 25 kΩ at 1 kHz. The Ti and Pt encapsulated ZnO NWs have the potential to offer an alternative microelectrode modality that could be attractive to electrochemical and biological sensing applications.

Insect odorant receptor-based biosensors: Current status and prospects.

Whilst the senses of vision and hearing have been successfully automated and miniaturized in portable formats (e.g. smart phone), this is yet to be achieved with the sense of smell. This is because the sensing challenge is not trivial as it involves navigating a chemosensory space comprising thousands of volatile organic compounds. Distinct aroma recognition is based on detecting unique combinations of volatile organic compounds. In natural olfactory systems this is accomplished by employing odorant receptors (ORs) with varying specificities, together with combinatorial neural coding mechanisms. Attempts to mimic the remarkable sensitivity and accuracy of natural olfactory systems has therefore been challenging. Current portable chemical sensors for odorant detection are neither sensitive nor selective, prompting research exploring artificial olfactory devices that use natural OR proteins for sensing. Much research activity to develop OR based biosensors has concentrated on mammalian ORs, however, insect ORs have not been explored as extensively. Insects possess an extraordinary sense of smell due to a repertoire of odorant receptors evolved to interpret olfactory cues vital to the insects' survival. The potential of insect ORs as sensing elements is only now being unlocked through recent research efforts to understand their structure, ligand binding mechanisms and development of odorant biosensors. Like their mammalian counterparts, there are many challenges with working with insect ORs. These include expression, purification and presentation of the insect OR in a stable display format compatible with an effective transduction methodology while maintaining OR structure and function. Despite these challenges, significant progress has been demonstrated in developing OR-based biosensors which exploit insect ORs in cells, lipid bilayers, liposomes and nanodisc formats. Ultrasensitive and highly selective detection of volatile organic compounds has been validated by coupling these insect OR display formats with transduction methodologies spanning optical (fluorescence) and electrical (field effect transistors, electrochemical impedance spectroscopy) techniques. This review summarizes the current status of insect OR based biosensors and their future outlook.

Comparison of Duplex and Quadruplex Folding Structure Adenosine Aptamers for Carbon Nanotube Field Effect Transistor Aptasensors.

Carbon nanotube field effect transistor (CNT FET) aptasensors have been investigated for the detection of adenosine using two different aptamer sequences, a 35-mer and a 27-mer. We found limits of detection for adenosine of 100 pM and 320 nM for the 35-mer and 27-mer aptamers, with dissociation constants of 1.2 nM and 160 nM, respectively. Upon analyte recognition the 35-mer adenosine aptamer adopts a compact G-quadruplex structure while the 27-mer adenosine aptamer changes to a folded duplex. Using the CNT FET aptasensor platform adenosine could be detected with high sensitivity over the range of 100 pM to 10 µM, highlighting the suitability of the CNT FET aptasensor platform for high performance adenosine detection. The aptamer restructuring format is critical for high sensitivity with the G-quadraplex aptasensor having a 130-fold smaller dissociation constant than the duplex forming aptasensor.

Investigation of Fractal Carbon Nanotube Networks for Biophilic Neural Sensing Applications.

We propose a carbon-nanotube-based neural sensor designed to exploit the electrical sensitivity of an inhomogeneous fractal network of conducting channels. This network forms the active layer of a multi-electrode field effect transistor that in future applications will be gated by the electrical potential associated with neuronal signals. Using a combination of simulated and fabricated networks, we show that thin films of randomly-arranged carbon nanotubes (CNTs) self-assemble into a network featuring statistical fractal characteristics. The extent to which the network's non-linear responses will generate a superior detection of the neuron's signal is expected to depend on both the CNT electrical properties and the geometric properties of the assembled network. We therefore perform exploratory experiments that use metallic gates to mimic the potentials generated by neurons. We demonstrate that the fractal scaling properties of the network, along with their intrinsic asymmetry, generate electrical signatures that depend on the potential's location. We discuss how these properties can be exploited for future neural sensors.

Directed self-assembly of peptide-diketopyrrolopyrrole conjugates - a platform for bio-organic thin film preparation.

Increased water solubility and long-range intermolecular ordering have been introduced into the fluorescent organic molecule thiophene-diketopyrrolopyrrole (TDPP) via its conjugation to the octapeptide HEFISTAH, which is derived from the protein-protein ß-interface of the homo-tetramer protein diaminopimelate decarboxylase. The octapeptide, and its TDPP mono- and cross-linked conjugates were synthesised using 9-fluorenylmethoxycarbonyl (Fmoc) based solid-phase peptide synthesis (SPPS). Unlike the unmodified peptide, the resulting mono-linked and cross-linked peptides showed a fibrous morphology and formed hydrogels at 4 wt% in water at neutral pH, but failed to assemble at pH 2 and pH 9. Further peptide characterization showed that the TDPP organic core enhances peptide self-assembly and that both peptides assembled into fibers with a parallel ß-sheet structure. Furthermore, UV-vis spectroscopic analysis suggests that the TDPP molecules form H-type aggregates where the chromophores are likely to be co-facially packed, but rotationally and/or laterally offset from one another. This intermolecular coupling indicates that π-π stacking interactions are highly likely - a favourable sign for charge transport. The enhanced aqueous solubility and self-assembling properties of the TDPP-peptide conjugates allowed the successful preparation of thin films. Atomic force microscopy, X-ray diffraction and UV-vis spectroscopic analysis of these thin films revealed that the hybrid materials retained a fibrous morphology, ß-sheet structures and strong intermolecular coupling between neighbouring TDPP molecules. These results open an exciting avenue for bio-organic materials development, through structural and electronic tuning of the TDPP core.

Biosensing with Insect Odorant Receptor Nanodiscs and Carbon Nanotube Field-Effect Transistors.

Insect odorant receptors have been reconstituted into lipid nanodiscs and tethered to carbon nanotube field-effect transistors to function as a biosensor. Here, four different insect odorant receptors (ORs) from Drosophila melanogaster (DmelOR10a, DmelOR22a, DmelOR35a, and DmelOR71a) were expressed in Sf9 cells, purified, and reconstituted into lipid nanodiscs. We have demonstrated that each of these ORs produce a selective and highly sensitive electrical response to their respective positive ligands, methyl salicylate, methyl hexanoate, trans-2-hexen-1-al, and 4-ethylguaiacol, with limits of detection in the low femtomolar range. No detection was observed for each OR against control ligands, and empty nanodiscs showed no specific sensor signal for any of the odorant molecules. Our results are the first evidence that insect ORs can be integrated into lipid nanodiscs and used as primary sensing elements for bioelectronic nose technologies.

Improved uniaxial dielectric properties in aligned diisopropylammonium bromide (DIPAB) doped poly(vinylidene difluoride) (PVDF) nanofibers.

Diisopropylammonium bromide (DIPAB) doped poly(vinylidene difluoride) (PVDF) nanofibers (5, 10 and 24 wt% DIPAB doping) with improved and tunable dielectric properties were synthesised via electrospinning. DIPAB nanoparticles were grown in situ during the nanofiber formation. X-Ray diffraction (XRD) patterns and Fourier transform infrared spectroscopy (FTIR) proved that electrospinning of DIPAB doped PVDF solutions led to the formation of a highly electro-active ß-phase in the nanofibers. Electrospinning in the presence of DIPAB inside PVDF led to very well aligned nanofibers with preferred (001) orientation that further enhanced the effective dipole moments in the nanofiber structures. The dielectric properties of the composite nanofibers were significantly enhanced due to the improved orientation, ionic and interfacial polarisation upon the applied electrospinning process, ionic nature of DIPAB and the interface between the PVDF nanofibers and equally dispersed DIPAB nanoparticles inside them, respectively. The relative dielectric constant of the PVDF nanofibers was improved from 8.5 to 102.5 when nanofibers were doped with 5% of DIPAB. Incorporating DIPAB in PVDF nanofibers has been shown to be an effective way to improve the structural and dielectric properties of PVDF.

Metallic-semiconducting junctions create sensing hot-spots in carbon nanotube FET aptasensors near percolation.

Easily fabricated random network carbon nanotube field-effect transistors (CNT-FETs) have benefitted from improved separation techniques to deliver CNTs with current formulations providing at least 99% semiconducting tube content. Amongst the most promising applications of this device platform are electronic biosensors, where the network conduction is affected through tethered probes such as aptamers which act as molecular scale electrostatic gates. However, the prevailing assumption that these biosensor devices would be optimized if metallic tubes were entirely eliminated has not been examined. Here, we show that metallic-semiconducting junctions in aptasensors are sensing hotspots and that their impact on sensing is heightened by the CNT network's proximity to percolation. First, we use a biased conducting AFM tip to gate a CNT-FET at the nanoscale and demonstrate that the strongest device response occurs when gating at metallic-semiconducting junctions. Second, we resolve the target sensitivity of an aptasensor as a function of tube density and show heightened sensitivity at densities close to the percolation threshold. We find the strongest sensing response where the 1% of metallic tubes generate a high density of metallic-semiconducting junctions but cannot form a percolated metallic path across the network. These findings highlight the critical role of metallic tubes in CNT-FET biosensor devices and demonstrate that network composition is an important variable to boost the performance of electronic biosensors.

Data on liquid gated CNT network FETs on flexible substrates.

This article presents the raw and analyzed data from a set of experiments performed to study the role of junctions on the electrostatic gating of carbon nanotube (CNT) network field effect transistor (FET) aptasensors. It consists of the raw data used for the calculation of junction and bundle densities and describes the calculation of metallic content of the bundles. In addition, the data set consists of the electrical measurement data in a liquid gated environment for 119 different devices with four different CNT densities and summarizes their electrical properties. The data presented in this article are related to research article titled "Metallic-semiconducting junctions create sensing hot-spots in carbon nanotube FET aptasensors near percolation" (doi:10.1016/j.bios.2018.09.021) [1].

Electrostatic gating in carbon nanotube aptasensors.

Synthetic DNA aptamer receptors could boost the prospects of carbon nanotube (CNT)-based electronic biosensors if signal transduction can be understood and engineered. Here, we report CNT aptasensors for potassium ions that clearly demonstrate aptamer-induced electrostatic gating of electronic conduction. The CNT network devices were fabricated on flexible substrates via a facile solution processing route and non-covalently functionalised with potassium binding aptamers. Monotonic increases in CNT conduction were observed in response to increasing potassium ion concentration, with a level of detection as low as 10 picomolar. The signal was shown to arise from a specific aptamer-target interaction that stabilises a G-quadruplex structure, bringing high negative charge density near the CNT channel. Electrostatic gating is established via the specificity and the sign of the current response, and by observing its suppression when higher ionic strength decreases the Debye length at the CNT-water interface. Sensitivity towards potassium and selectivity against other ions is demonstrated in both resistive mode and real time transistor mode measurements. The effective device architecture presented, along with the identification of clear response signatures, should inform the development of new electronic biosensors using the growing library of aptamer receptors.

Direct measurement of charge transport through helical poly(ethyl propiolate) nanorods wired into gaps in single walled carbon nanotubes.

We report the direct measurement of electrical transport through rod-like polymer molecules, of poly(ethyl propiolate) (PEP), utilizing single walled carbon nanotubes (SWNTs) as electrodes. The electrical properties of the devices were measured (i) before cutting a SWNT, (ii) when a SWNT was cut and (iii) after PEP deposition into the nanoscale gap in a cut SWNT. The gate-dependent electrical properties showed a reduction in current from I(on) = 2.4 x 10(-7) A for SWNT devices to I(on) = 3.6 x 10(-9) A for PEP bridge devices, both with the ON/OFF ratio of 10(4). Similarly, metallic SWNT devices showed a reduction in current from a few hundreds of microA for a SWNT device to a few nA for a PEP-SWNT structure. The current density of a single PEP molecule is 10(5)-10(6) A cm(-2), which is relatively high, indicating that the PEP molecule can carry significant current. Use of SWNT electrodes was seen to be an effective method of contacting PEP nanorods to facilitate electrical measurements.

Electronic properties of n-type carbon nanotubes prepared by CF4 plasma fluorination and amino functionalization.

Single-walled carbon nanotubes (SWNTs) have been fluorinated by CF4 plasma exposure and further functionalized with 1,2-diaminoethane. The degree of amino functionalization is dependent on the degree of initial fluorination rather than oxygen or carbon defects. Reaction at both ends of 1,2-diaminoethane was observed to increase with fluorine content. Back-gated SWNT devices have shown p-type semiconducting behavior for CF4-functionalized SWNTs and n-type semiconducting behavior for amino-functionalized SWNTs. The degree of n-type behavior increases with the amount of nitrogen attached to the SWNTs.