Species identification and strain discrimination of fermentation yeasts Saccharomyces cerevisiae and Saccharomyces uvarum using Raman spectroscopy and convolutional neural networks.

IMPORTANCE: The use of S. cerevisiae and S. uvarum yeast starter cultures is a common practice in the alcoholic beverage fermentation industry. As yeast strains from different or the same species have variable fermentation properties, rapid and reliable typing of yeast strains plays an important role in the final quality of the product. In this study, Raman spectroscopy combined with CNN achieved accurate identification of S. cerevisiae and S. uvarum isolates at both the species and strain levels in a rapid, non-destructive, and easy-to-operate manner. This approach can be utilized to test the identity of commercialized dry yeast products and to monitor the diversity of yeast strains during fermentation. It provides great benefits as a high-throughput screening method for agri-food and the alcoholic beverage fermentation industry. This proposed method has the potential to be a powerful tool to discriminate S. cerevisiae and S. uvarum strains in taxonomic, ecological studies and fermentation applications.

Infrared spectra of the SARS-CoV-2 spike receptor-binding domain: Molecular dynamics simulations.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread around the world rapidly, which seriously threatens to human health and safety. The rapid detection of the virus in the early stage is very important to prevent the cross infection and transmission. It is also a key link in the post-treatment examination. This paper has explored the infrared (IR) spectra of spike protein receptor-binding domain (RBD) for SARS-CoV-2 using molecular dynamics simulations, and the absorption bands are assigned. The calculated IR spectra of water and insulin are compared with that measured in the related literatures. The results showed that O-H stretching vibration generated a strong absorption band located around 3591 cm-1, the oscillator strength of 310 K is slightly higher than that at 298 K. The absorption peaks have a small red shift or blue shift with the change of temperature. As a theoretical basis for the optical detection of SARS-CoV-2 virus, this work will play a positive role in promoting the development of new virus detection technology.

Conditional Generative Adversarial Network for Spectral Recovery to Accelerate Single-Cell Raman Spectroscopic Analysis.

Raman spectroscopy is a powerful tool to investigate cellular heterogeneity. However, Raman spectra for single-cell analysis are hindered by a low signal-to-noise ratio (SNR). Here, we demonstrate a simple and reliable spectral recovery conditional generative adversarial network (SRGAN). SRGAN reduced the data acquisition time by 1 order of magnitude (i.e., 30 vs 3 s) by improving the SNR by a factor of ∼6. We classified five major foodborne bacteria based on single-cell Raman spectra to further evaluate the performance of SRGAN. Spectra processed using SRGAN achieved an identification accuracy of 94.9%, compared to 60.5% using unprocessed Raman spectra. SRGAN can accelerate spectral collection to improve the throughput of Raman spectroscopy and enable real-time monitoring of single living cells.

Insights into the conformation changes of SARS-CoV-2 spike receptor-binding domain on graphene.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been widely spread in the world, causing more than two million deaths and seriously threatening human life. Effective protection measures are important to prevent the infection and spreading of the virus. To explore the effects of graphene on the virus adsorption and its biological properties, the adsorption process of the receptor binding domain (RBD) of SARS-CoV-2 on graphene has been investigated by molecular dynamics simulations in this paper. The results show that RBD can be quickly adsorbed onto the surface of graphene due to π - π stacking and hydrophobic interactions. Residue PHE486 with benzene ring has stronger adsorption force and the maximum contact area with graphene. Graphene significantly affects the secondary structure of RBD area, especially on the three key sites of binding with human ACE2, GLY476, PHE486 and ASN487. The binding free energy of RBD and graphene shows that the adsorption is irreversible. Undoubtedly, these changes will inevitably affect the pathogenicity of the virus. Therefore, this study provides a theoretical basis for the application of graphene in the protection of SARS-CoV-2, and also provides a reference for the potential application of graphene in the biomedical field.

On-chip monolithic Fourier transform spectrometers assisted by cGAN spectral prediction.

Silicon photonic spatial heterodyne Fourier transform spectrometers (SH-FTSs) are attractive with chip-scale monolithic arrays of imbalanced Mach-Zehnder interferometers; however, there exist optical path difference (OPD) errors from the inevitable fabrication imperfection, which will severely distort the retrieved spectra. In this Letter, we propose that a predictive model can be created for rapid and accurate spectral recovery based on the conditional generative adversarial network (cGAN) featuring strong input-on-output supervision, instead of both complicated physical OPD modification and time-consuming iterative spectral calculation. As a demonstration, cGAN spectral prediction was performed for our previously presented dual-polarized SH-FTS with large OPD errors [Opt. Lett.44, 2923 (2019)OPLEDP0146-959210.1364/OL.44.002923]. Due to the strong noise-resistant capability, the cGAN-predicted spectra can stay reliable, even though the signal-to-noise ratio of acquired interferograms dramatically drops from 1000 to 100, implying a lower limit of detection.

Arcobacter Identification and Species Determination Using Raman Spectroscopy Combined with Neural Networks.

Rapid and accurate identification of Arcobacter is of great importance because it is considered an emerging food- and waterborne pathogen and potential zoonotic agent. Raman spectroscopy can differentiate bacteria based on Raman scattering spectral patterns of whole cells in a fast, reagentless, and easy-to-use manner. We aimed to detect and discriminate Arcobacter bacteria at the species level using confocal micro-Raman spectroscopy (785 nm) coupled with neural networks. A total of 82 reference and field isolates of 18 Arcobacter species from clinical, environmental, and agri-food sources were included. We determined that the bacterial cultivation time and growth temperature did not significantly influence the Raman spectral reproducibility and discrimination capability. The genus Arcobacter could be successfully differentiated from the closely related genera Campylobacter and Helicobacter using principal-component analysis. For the identification of Arcobacter to the species level, an accuracy of 97.2% was achieved for all 18 Arcobacter species using Raman spectroscopy combined with a convolutional neural network (CNN). The predictive capability of Raman-CNN was further validated using an independent data set of 12 Arcobacter strains. Furthermore, a Raman spectroscopy-based fully connected artificial neural network (ANN) was constructed to determine the actual ratio of a specific Arcobacter species in a bacterial mixture ranging from 5% to 100% by biomass (regression coefficient >0.99). The application of both CNN and fully connected ANN improved the accuracy of Raman spectroscopy for bacterial species determination compared to the conventional chemometrics. This newly developed approach enables rapid identification and species determination of Arcobacter within an hour following cultivation.IMPORTANCE Rapid identification of bacterial pathogens is critical for developing an early warning system and performing epidemiological investigation. Arcobacter is an emerging foodborne pathogen and has become more important in recent decades. The incidence of Arcobacter species in the agro-ecosystem is probably underestimated mainly due to the limitation in the available detection and characterization techniques. Raman spectroscopy combined with machine learning can accurately identify Arcobacter at the species level in a rapid and reliable manner, providing a promising tool for epidemiological surveillance of this microbe in the agri-food chain. The knowledge elicited from this study has the potential to be used for routine bacterial screening and diagnostics by the government, food industry, and clinics.

Speeding up Raman spectral imaging by the three-dimensional low rank estimation method.

Raman spectral imaging has been widely used as a very important analytical tool in various fields. For obtaining the high spectral signal-to-noise ratio Raman images, the long integration time is necessary, which is placing a limit on the application of Raman spectral imaging. We introduce a simple and feasible numerical method of the Three-dimensional Low Rank Estimation (3D-LRE), which can speed up the data acquisition process of the Raman spectral imaging. The spectral signal-to-noise ratio of the Raman images can be increased by over 75 times and the speed of the data acquisition can be improved by over 30 times. By combining with line-scan or multifocus-scan techniques, the Raman images can be obtained in a few seconds.

High-resolution broadband sum frequency generation vibrational spectroscopy using intrapulse interference.

We theoretically propose a new approach to obtain high-spectral-resolution sum frequency generation (SFG) vibrational spectra using intrapulse interference. By introducing a π-step phase modulation to the broadband 800 nm pulse, the broadband 800 nm laser pulse splits into two distinguishable pulses in the time domain with a fixed time delay. The resolution of the intrapulse interference SFG can be better than 1 cm-1 and is limited only by the spectral resolution of the spectrometer. This approach can accurately retrieve the amplitude and the relative phase of vibrational peaks. Additionally, the sensitivity of SFG is enhanced by adopting femtosecond IR and femtosecond visible pulses.

Enhancing the signal-to-noise ratio of FTIR spectrometers by a digital J-Stop.

Fourier transform infrared (FTIR) spectrometers have been widely used as very important analytical tools in various fields. Owing to the Jacquinot Stop (J-Stop), high throughput is a widely recognized advantage inherent to Fourier transform interferometers. However, there is a fundamental trade-off between the throughput and spectral resolution, which is primarily affected by the size of the J-Stop. So far, no effective optimization methods have been provided to break the trade-off. In this paper, we introduce a numeric technique of the digital J-Stop, which has been experimentally validated using the FTIR spectra collected from a commercial spectrometer. The result shows that the throughput can be increased by ~12 times, while the spectral resolution is also improved. In this way, the signal-to-noise ratio (SNR) gets improved by ~3 times.

Improving the resolution and the throughput of spectrometers by a digital projection slit.

The contradiction between spectral resolution and throughput for the optical spectrometers is still a problem that needs to be solved. We introduce a simple and feasible method of the digital projection slit (DPS), which can improve both the spectral resolution and throughput of the optical spectrometer. The DPS spectrum is accurate and reliable without using the optical transfer function (OTF) of the optical spectrometer. The method has been successfully applied in the fiber spectrometer and the Raman spectrometer. The resolution of the recovered spectra can be increased by ~3 times and the throughput can be increased by ~5 times.