Abstract
RNA sequencing libraries are typically processed before sequencing. We compared the performance of automated mRNA enrichment (autoE) and manual mRNA enrichment (manE) with globin depletion, and automated rRNA depletion (autoD) with globin depletion prior to sequencing. PAXgene tubes from 21 people living with HIV (PLWH) and 7 people without HIV were used, with RT-qPCR data from unprocessed RNA as an orthogonal comparator. Duplicate reads were higher for manE (58–61%) and autoE (59–70%) than autoD (29–33%). AutoE had the highest mapping to exonic regions (72–78% vs. 50–74% manE and 30–39% autoD). Globin transcripts were more abundant for autoD than autoE and manE, but remained low (median 10% vs. 0.1 and 1.3%). 94% of protein-coding genes were detected by all methods vs. 63–66% for other RNA-biotypes. All methods correlated strongly with qPCR data for highly expressed genes but not lowly expressed genes; concordance was lowest for autoE. When comparing PLWH with controls, manE and autoD identified more differentially expressed genes than autoE. While all methods exceeded the minimum threshold required for downstream analysis, autoD provided the most stable sequencing quality. In conclusion, manE and autoD yielded similar results with autoD providing advantages of robotic liquid handling for large-scale studies. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1038/s41598-025-32961-4.