AI Applications for COVID-19 Research and Other News

May 22, 2020

May 22, 2020 | A pair of studies demonstrate antibody responses to COVID-19 in primates. Researchers are applying artificial intelligence to lung images of COVID-19 patients. An accurate, easy-to-use, fully disposable, rapid and handheld test that consumers and healthcare providers in clinics can use to detect active SARS-CoV-2. We round up the week’s research and industry news for COVID-19.

Literature Updates

A pair of studies led by researchers at Beth Israel Deaconess Medical Center (BIDMC) were published in the journal Science. In the first study, the team demonstrated that six candidate DNA vaccines induced neutralizing antibody responses and protected against SARS-CoV-2 in rhesus macaques. DOI: 10.1126/science.abc6284. In the second study, the team demonstrated that macaques that recovered from COVID-19 developed natural protective immunity against re-infection. Upon second exposure, the animals demonstrated near-complete protection against the virus. These data suggest natural protective immunity against COVID-19 in this model. DOI: 10.1126/science.abc4776.

Insilico Medicine has co-authored a preprint paper with describing 10 potential small molecule inhibitors targeting the SARS-CoV-2 main protease that were generated by artificial intelligence (AI). The approach taken, called “AI imagination”, involves teaching a computer what the target protein looks like, letting it “sniff out” the binding pockets, and then “imagine” molecules with certain features having to do with the likelihood that a molecule will fit inside, and stay inside, a binding pocket. The AI-generated molecules are some of the first non-covalent drug candidates for COVID-19; non-covalent inhibitors are safer and more selective than covalent inhibitors. DOI: 10.13140/RG.2.2.13846.98881.

Researchers from the La Jolla Institute for Immunology have published the first cellular immunology data to help guide social distancing recommendations. Published in Cell, the study documents a robust antiviral immune response to SARS-CoV-2 in a group of 20 adults who had recovered from COVID-19. The findings show that the body's immune system is able to recognize SARS-CoV-2 in many ways, dispelling fears that the virus may elude ongoing efforts to create an effective vaccine. DOI: 10.1016/j.cell.2020.05.015.

Previous data from COVID-19 patients suggests that cigarette smokers are more likely to have health complications. One possible reason, researchers report in Developmental Cell, is that smoking increases the gene expression of ACE2—the protein that binds SARS-CoV-2—which may promote COVID-19 infection. The study suggests that prolonged smoking could cause an increase of the ACE2 protein in the lungs, possibly resulting in a higher rate of morbidity in patients. ACE2, or Angiotensin Converting Enzyme 2, is a regulatory protein that has been linked to vulnerability to the 2003 SARS (2003) virus. DOI: 10.1016/j.devcel.2020.05.

Researchers from the University of Trento have published a novel fully-annotated dataset of lung ultrasonography (LUS) images from COVID-19 patients collected from several Italian hospitals, with labels indicating the degree of disease severity at a frame-level, video-level, and pixel-level (segmentation masks). Leveraging these data, they introduce a novel deep network, derived from Spatial Transformer Networks, which simultaneously predicts the disease severity score associated to an input frame and provides localization of pathological artefacts in a weakly-supervised way. They also benchmark state of the art deep models for estimating pixel-level segmentations of COVID-19 imaging biomarkers. Results were published in IEEE Transactions on Medical Imaging (Early Access). DOI: 10.1109/TMI.2020.2994459.

Leading immunologists in Japan reviewed two recent studies by Zhou et al. and Hoffmann et al. in order to understand their implications for finding effective therapeutic strategies for ARDS in COVID-19 patients. Based on their review, they are proposing a possible molecular mechanism that causes the massive release of proinflammatory cytokines, or a cytokine storm, that leads to acute respiratory distress syndrome (ARDS) in COVID-19 patients. Their suggestions are published in the journal Immunity. DOI: 10.1016/j.immuni.2020.04.003.

Mount Sinai researchers are also applying artificial intelligence to lung images of COVID-19 patients. The Mount Sinai team has developed an algorithm that can rapidly detect COVID-19 based on how lung disease looks in computed tomography (CT scans) of the chest, in combination with patient information including symptoms, age, bloodwork, and possible contact with someone infected with the virus. This study, published in Nature Medicine, could help hospitals across the world quickly detect the virus, isolate patients, and prevent it from spreading during this pandemic. DOI: 10.1038/s41591-020-0931-3.

Patients infected with either severe acute respiratory syndrome coronavirus (SARS-CoV) or SARS-CoV-2 produce antibodies that bind to the other coronavirus, but the cross-reactive antibodies are not cross protective, at least in cell-culture experiments, researchers report in Cell Reports. Researchers analyzed blood samples collected from 15 SARS-CoV-2-infected patients in Hong Kong between 2 and 22 days after the onset of symptoms. Compared to blood samples from healthy controls, the five samples collected from patients 11 days after symptom onset or later had antibodies capable of binding to the RBD and other parts of the S protein on both SARS-CoV-2 and SARS-CoV. The researchers also analyzed blood samples collected from seven patients 3 to 6 months after infection with SARS-CoV. Compared to blood samples from healthy controls, those collected from patients had antibodies capable of binding to the RBD and other parts of the S protein on SARS-CoV-2. It remains unclear whether such antibodies offer cross protection in the human body or potentiate disease. DOI: 10.1016/j.celrep.2020.

An antibody called S309, first identified in a blood sample from a patient who recovered from Severe Acute Respiratory Syndrome in 2003, inhibits related coronaviruses, including SARS-CoV-2. The findings have been accepted for publication in Nature and the antibody is now on a fast-track development and testing path at Vir Biotechnology. It has not yet been shown to be protective in living systems. The S309 antibody is particularly potent at targeting and disabling the spike protein that promotes the coronavirus entry into cells. It was able to neutralize SARS CoV-2 by engaging with a section of the spike protein nearby the attachment site to the host cell. DOI: 10.1038/s41586-020-2349-y.

Industry Updates

Illumina and the Illumina Corporate Foundation have committed more than $10M to COVID-19-related research and support. To support public health efforts associated with SARS-CoV-2 surveillance, sequencing and monitoring, Illumina is providing in kind donations valued at approximately $5M dollars for instruments and consumables. In addition, we dedicated more than $2M to COVID-19-related research efforts. To advance scientific research, Illumina, Inc. recently released the SARS-CoV-2 Data Toolkit, making it easier for researchers to detect and identify the viral sequence in their samples and contribute their findings to critical public databases. The toolkit is freely accessible on BaseSpace Sequence Hub until October 2020. Finally, the Illumina Foundation has awarded more than $3M in philanthropic donations, of which $1M has gone to the CDC Foundation COVID Emergency Response Fund, and $2M to local community-based initiatives in the areas where we operate. Press release.

Mammoth Biosciences and GSK Consumer Healthcare are collaborating to develop an accurate, easy-to-use, fully disposable, rapid and handheld test that consumers and healthcare providers in clinics can use to detect active SARS-CoV-2. The collaboration will use Mammoth Biosciences’ CRISPR-based DETECTR platform that can identify and signal the presence of viral RNA strands obtained through a simple nasal swab. The test, which has the potential to deliver point-of-use results in less than 20 minutes, will be available first in the clinical setting and ultimately by consumers at home. Press release.

iSpecimen is partnering with myOnsite HealthCare to offer mobile phlebotomy support that will bring convenient, at-home sample collection to patients and donors interested in advancing research who would otherwise have limited ability to participate. The mobile service extends iSpecimen’s ongoing mission to advance medical research by efficiently connecting researchers with patients, biospecimens, and the data they need to perform their important work. It also expands the company’s recent initiatives to identify, collect, aggregate, manage, and ship specimens from COVID-19 patients. Press release.