Comprehensive and bias-aware analysis of small RNA NGS data for biomarker detection with caRNAge

Abstract

Aim: Small RNA sequencing (small RNA-Seq) is a widely used and important method for biomarker discovery in extracellular vesicles (EVs), which carry small RNAs among their cargo and protect them from degradation in liquid biopsies. However, existing computational tools for small RNA-Seq data are not tailored to EV-RNA. Methods: We present comprehensive and bias-aware analysis of small RNA gene expression (caRNAge), a biomarker discovery pipeline for small RNA-Seq data intended for users without extensive bioinformatic expertise. Ready to use as a Snakemake pipeline, caRNAge assists users in bias-aware normalization, differential gene expression analysis, and pathway over-representation analysis. Results: Possible confounding factors can be identified and corrected to maximize data comparability between experiments and biomarker robustness. caRNAge is not limited to miRNAs; it also supports the detection of miRNA isomer (isomiRs), tRNA-derived fragments (tRFs), transfer RNA (tRNAs), and other small RNA species, thereby enhancing the specificity of biomarker signatures. Specific tissues can be selected for target prediction and pathway analysis to eliminate false-positive results. All results are provided in accessible and interactive HTML reports. We provide example data for each step of the pipeline to demonstrate caRNAge’s capabilities and advantages across three EV-miRNA biomarker discovery use cases. The pipeline is not limited to EV-miRNA data but can be used for data evaluation in any small RNA-Seq experiment. Conclusion: By assisting users at every step of the discovery workflow, caRNAge fills an important gap in the field of EV-associated transcriptomic biomarker development. The pipeline is freely available at https://gitlab.lrz.de/kirchner/carnage.