Advanced Technology Development Centre, Indian Institute of Technology Kharagpur, Kharagpur 721302, India.
Department of Electronics and Electrical Communication Engineering, Indian Institute of Technology Kharagpur, Kharagpur 721302, India.
Anal Chem. 2024 Sep 17;96(37):14749-14758. doi: 10.1021/acs.analchem.4c01260. Epub 2024 Aug 31.
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has triggered an ongoing global pandemic, necessitating rapid and accurate diagnostic tools to monitor emerging variants and preparedness for the next outbreak. This study introduces a multidisciplinary approach combining Fourier Transform Infrared (FTIR) microspectroscopy and Machine learning to comprehensively characterize and strain-type SARS-CoV-2 variants. FTIR analysis of pharyngeal swabs from different pandemic waves revealed distinct vibrational profiles, particularly in nucleic acid and protein vibrations. The spectral wavenumber range between 1150 and 1240 cm was identified as the classification marker, distinguishing Healthy (noninfected) and infected samples. Machine learning algorithms, with neural networks exhibiting superior performance, successfully classified SARS-CoV-2 variants with a remarkable accuracy of 98.6%. Neural networks were also able to identify and differentiate a small cohort infected with influenza A variants, H1N1 and H3N2, from SARS-CoV-2-infected and Healthy samples. FTIR measurements further show distinct red shifts in vibrational energy and secondary structural alterations in the spike proteins of more transmissible forms of SARS-CoV-2 variants, providing experimental validation of the computational data. This integrated approach presents a promising avenue for rapid and reliable SARS-CoV-2 variant identification, enhancing our understanding of viral evolution and aiding in diagnostic advancements, particularly for an infectious disease with unknown etiology.
严重急性呼吸综合征冠状病毒 2（SARS-CoV-2）引发了持续的全球大流行，因此需要快速、准确的诊断工具来监测新出现的变异株，并为下一次疫情做好准备。本研究采用傅里叶变换红外（FTIR）显微光谱学和机器学习相结合的多学科方法，全面表征和分型 SARS-CoV-2 变异株。对来自不同大流行波次的咽拭子进行 FTIR 分析，揭示了独特的振动特征，特别是在核酸和蛋白质振动方面。在 1150 到 1240 厘米的光谱波数范围内，发现了可用于区分健康（未感染）和感染样本的分类标志物。机器学习算法，尤其是神经网络算法，成功地对 SARS-CoV-2 变异株进行了分类，准确率达到了 98.6%。神经网络还能够识别和区分一小部分感染甲型流感 H1N1 和 H3N2 变异株的样本，以及 SARS-CoV-2 感染和健康样本。FTIR 测量进一步显示，传染性更强的 SARS-CoV-2 变异株的刺突蛋白中振动能量和二级结构发生了明显的红移，为计算数据提供了实验验证。这种综合方法为快速、可靠地鉴定 SARS-CoV-2 变异株提供了一种很有前途的途径，有助于加深我们对病毒进化的理解，并为诊断技术的发展提供支持，尤其是对于病因不明的传染病。
