Molecular landscape of respiratory infection: A large-scale, multi-centre blood transcriptome dataset

Abstract

Respiratory infections pose significant challenges to global health, impacting millions of individuals annually. Understanding the molecular mechanisms underlying the pathogenicity of these infections is crucial for developing effective interventions. RNA sequencing provides insights into a patient's global transcriptome changes, facilitating the identification of host gene signatures in response to infection and potential therapeutic targets. Here we present an extensive whole blood transcriptome dataset from a demographically diverse cohort of 502 patients with infections including COVID-19, seasonal coronavirus, influenza A or influenza B, sepsis, septic shock, and co-infections (Viral/Viral, Bacterial/Viral, Bacterial/Viral/Fungal, Viral/Fungal, Viral/ Viral/Fungal). The cohort size and depth of data showcase its potential to unravel respiratory infection pathogenesis for the development of better diagnostics, treatments, and preventive strategies for respiratory infections and future global health crises.

Fig. 1

Experimental flow chart outlining the study design, from sample collection to data sharing. This diagram illustrates the key steps involved in the study, including participant recruitment, sample processing, data analysis, and data deposition into public repositories for sharing.

Fig. 2

Raw and analysis-ready data quality after mapping and controlling unwanted batch effects. (a) The principal components analysis plot is grouped by disease status. (b) DESeq 2 generated dispersion parameter estimates for each gene. (c) Mean quality scores of raw sequencing data per FASTQC file across the 150 bp read length were generated with FastQC and MultiQC. All bases across reads had a mean quality score of Phred 30 or greater, except at position one, where the quality was a minimum of Phred 16 (97.5% accuracy).