In Vivo Cluster AI Prediction (CLAIRE) of COVID-19 Disease Progression

点击翻译按钮获取中文摘要

ABSTRACT The coronavirus COVID-19 pandemic, which early this year forced entire countries into lockdown, has reached a global death toll of 890,000+ by early September 2020. Based on the high number of COVID-19 cases that are asymptomatic but infectious, an estimated reproductive rate of infection of about 2 and a high mutation rate, it is expected that the virus will remain in the population as the influenza virus does. For hospitals serving areas whose economy relies on international travel, tourism, and cruise ship tourism, such as Miami-Dade county, new COVID-19 cases related to travel will require treatment during infection outbreaks which will strain health systems, especially during the infectious respiratory disease season in the winter. Patient risk factors during the current COVID-19 outbreak as well as during other viral outbreaks, such as seasonal influenza, are poorly characterized, consequently negatively affecting patient care. The saliva microbiome, which includes viruses and bacteria, is not currently used as in diagnostic tools. However, it may reveal risk factors associated with severe disease and/or a fatal outcome, and it allows for the detection and study of the viral RNA sequence for potential contact tracing and molecular epidemiology, all of which affect both vaccine and antiviral efficacy. In this proposed study, Lifetime Omics will develop CLAIRE, a proof-of-concept in vivo cluster AI platform for predicting disease progression of viral infectious respiratory diseases such as COVID-19 through the analysis of the saliva metagenome. The University of Miami Medical Group Infection Control (UMMGIC) division will collaborate in this effort by collecting saliva samples from COVID-19 patients with de-identified clinical information. The samples will undergo metagenomic sequencing and Lifetime Omics will repurpose algorithms used for prediction of in vivo HIV evolution to perform genetic/phylogenetic analysis on SARS-CoV-2 RNA sequences, estimating mutation rate and immune selection pressures and identifying both the in vivo quasispecies clusters and the geographic cluster to which the patient belongs. The CLAIRE models will be trained with public datasets and tested on the metagenomic sequences generated from saliva samples of UMMGIC patients with the goal of assisting physicians in predicting disease progression in COVID-19.