Machine learning analysis of single-cell gene regulatory networks reveals potential mechanisms of action of antimalarials against SARS-CoV-2

The efficiency of antimalarials chloroquine (CQ) and hydroxychloroquine (HCQ) in the prevention and treatment of coronavirus disease 2019 (COVID-19) is under intense debate. The mechanisms of action of antimalarials against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have not been fully elucidated. Here, we applied a network-based comparative analysis, implemented in our machine learning workflow—scTenifoldNet, to scRNA-seq data from COVID-19 patients with different levels of severity. scTenifoldNet is built upon principal component regression, low-rank tensor approximation, and manifold alignment, to construct and compare single-cell gene regulatory networks. Using the expression data in macrophages, we found that the expression of genes in the Malaria pathway is significantly differentially regulated between COVID-19 patients with moderate and severe symptoms. A significant difference in gene expression regulation patterns was also found for genes in the Phagosome pathway. These results help reveal mechanisms of action of CQ and HCQ during SARS-CoV-2 infection, which may involve cellular immunity modulation and lysosomal activity inhibition. Our findings present new evidence for the use of antimalarial drugs in COVID-19 treatment, especially for patients who are mildly affected or in the early stage of infection with SARS-CoV-2.