Transcriptome from Paired Samples Improves the Power of Comprehensive COVID-19 Host-Viral Characterization

Abstract

Previous transcriptome profiling studies showed significantly upregulated genes and altered biological pathways in acute COVID-19. However, changes in the transcriptional signatures during a defined time frame are not yet examined and described. The aims of this study included viral metagenomics and evaluation of the total expression in time-matched and tissue-matched paired COVID-19 samples with the analysis of the host splicing profile to reveal potential therapeutic targets. Prospective analysis of paired nasopharyngeal swabs (NPS) and blood (BL) samples from 18 COVID-19 patients with acute and resolved infection performed using Kallisto, Suppa2, Centrifuge, EdgeR, PantherDB, and L1000CDS2 tools. In NPS, we discovered 6 genes with changed splicing and 40 differentially expressed genes (DEG) that yielded 88 altered pathways. Blood samples yielded 15 alternatively spliced genes. Although the unpaired DEG analysis failed, pairing identified 78 genes and 242 altered pathways with meaningful clinical interpretation and new candidate drug combinations with up to 65% overlap. Metagenomics analyses showed SARS-CoV-2 dominance during and after the acute infection, with a significant reduction in NPS (0.008 vs. 0.002, p = 0.019). Even though both NPS and BL give meaningful insights into expression changes, this is the first demonstration of how the power of blood analysis is vastly maximized by pairing. The obtained results essentially showed that pairing is a determinant between a failed and a comprehensive study. Finally, the bioinformatics results prove to be a comprehensive tool for full-action insights, drug development, and infectious disease research when designed properly.

Keywords: COVID-19; RNA; SARS-CoV-2; differential expression (DEG); metagenomics; pairing; therapeutics; transcriptome.

Figure 1

String database protein–protein interactions. The nodes represent genes, while the line width represents the strength of interaction. There are three clusters of genes without interconnections—nineteen histone genes, an immunity-related cluster with five genes, and the smallest cluster with two interacting genes.

Figure 2

Hierarchically clustered heatmap of overlaps for individual perturbagens. Most substances match the genes from the immunoglobulin superfamily. Red genes are upregulated, and they are paired with blue perturbagen signatures of opposite sign. Opposite logic is valid for the downregulated PDZK1IP1 gene.