Fig. 1

Differential overexpression of HERV-K transcripts in the lower respiratory tract of critically ill COVID-19 patients is associated with early mortality. RNA sequencing of tracheal aspirates (TA) from severe cases (Supplementary Table 1) and nasopharyngeal swabs (NS) from mild cases¹⁷ was performed on the MGI-2000 RNA-seq platform, and high-quality sequences (Q ≥ 30) were selected for downstream analysis. A Percentage of virus-related reads in the mapped virome from the TA of severe COVID-19 patients, from the NS from COVID-19 mild cases, and from non-COVID TA Sequence Read Archive (SRA) (# SRX4213540, SRX4213544, SRX4213548, SRX4213551, SRX4213553, SRX3934905, SRX3934906, SRX3934910, and SRX3934932). B The percentage of HERV-K-related reads in the mapped virome from the TA of discharged and deceased severe COVID-19 patients compared to NS and non-COVID TA (# SRX4213540, SRX4213544, SRX4213548, SRX4213551, SRX4213553, SRX3934905, SRX3934906, SRX3934910, and SRX3934932). C Logistic regression analysis between HERV-K expression and odds of early (< 14 days) mortality in deceased COVID-19 patients. Red dotted lines represent the 95% CI, while black dotted lines mark the intersection where data in x-axis represent 0.5 (50%) probability. Insert receiver operating characteristic (ROC) curve for the prediction of early (< 14 days) mortality in deceased COVID-19 patients based on HERV-K expression. D HERV-K expression in TA over time (days) from ICU admission to death. E Heatmap of absolute HERV-K read counts for TA from severe COVID-19 patients, NS from mild cases and for the non-COVID TA with HERV-K presence. **= p < 0.01