Arya Sukumar, College of Agriculture, Vellayani
Cell biology has tremendously grown during the latter years of the past century due to the convergence of numerous techniques that vastly enhanced the field of microscopy. The application of Artificial Intelligence (AI) in life science has opened the door to a new era in diagnostic medicine. Urs Greber’s research group at the Department of Molecular Life Sciences, University of Zurich’s (UZH) has created an artificial neural network that can reliably detect cells infected with adenoviruses or herpes viruses. Adenoviruses are non-enveloped double-stranded DNA viruses with an icosahedral nucleocapsid. This virus infects the lining of the eyes, airways and lungs, intestine, urinary tract, and nervous system in humans. Herpesviruses have double-stranded DNA genomes that are relatively complicated and have icosapentahedral capsid made up of capsomers. Herpes viruses can infect the skin and nervous system cells. These viral infections result in a diverse set of cellular phenotypes with intricate biochemistry. Generally, the immune system inhibits the spread of these viruses, so no new viral particles are produced. But these same viruses can produce abrupt, severe infections in which infected cells release huge quantities of virus, allowing the infection to spread quickly. This can result in the severe acute lung or nervous system disorders.
Automatic detection of virus-infected cells
The use of live-cell imaging (LCI) to investigate viral infections is an important area with immense potential for breakthroughs in computational cell biology. Fluorescence microscopy is an important tool in life science research. For the first time, the researchers have demonstrated that a machine-learning system can identify cells with herpes or adenoviruses purely based on the fluorescence of the infected cell nucleus. The analytical approach combines fluorescence microscopy in living cells with deep-learning processes. Herpesviruses and adenoviruses modify the nucleic acid content of infected cells, which may be seen with a fluorescence microscope. Artificial neural networks (ANN) have recently improved artificial intelligence applications for cellular pattern identification in the biological sciences. The artificial neural network detects virulent viral infections in cells and also reveals how human cells respond to other viruses and bacteria. The network is trained to recognize patterns that can distinguish infected and uninfected cells using a large set of microscope images. The neural network automatically detects virus-infected cells after the training and validations are completed. With about 95 percent accuracy the algorithm can detect acute and severe infections. It is believed that several cellular mechanisms determine whether a cell is infected or not. The researchers have already identified that the nucleus’s internal pressure being higher during virulent infections than during chronic infections. Furthermore, viral proteins accumulate more quickly in the cells with lytic infection. Images of live cells with persistent infections in which virus particles proliferate quickly and the disintegrated cells were used as reference material by the artificial neural network. Using this procedure early detection of severe acute infections is also possible. The technique opens up new avenues for better understanding infections and the discovery of novel antiviral or antibacterial medicines.
Also read: Targeting viral protein ORF3a to fight Covid-19!
Reference:
Vardan Andriasyan, Artur Yakimovich, Anthony Petkidis, Fanny Georgi, Robert Witte, Daniel Puntener, Urs F. Greber. Microscopy deep learning predicts virus infections and reveals mechanics of lytic-infected cells. iScience, 2021; 24 (6): 102543 DOI: 10.1016/j.isci.2021.102543
One thought on “AI and fluorescent microscopy combo to predict viral infections”