Detecting unannotated splicing events in short-read RNA-seq with SAMI, a UMI-aware Nextflow pipeline.

Abstract

Although RNA-sequencing has replaced microarrays for gene expression profiling over the past 15 years, its full potential for splicing analysis in clinical settings remains underexploited. Most available tools are tailored for large cohorts or known isoforms, limiting their applicability in routine diagnostics where non-recurring events must be identified in low-dimension datasets. We present SAMI (Splicing Analysis with Molecular Indexes), a fully-integrated UMI-aware pipeline designed to detect splicing events diverging from transcript annotations. Building upon the well-proven STAR aligner, SAMI introduces original post-processing of gaps and potential intron retentions to maximize accuracy, along with clear graphical representations and tunable filtering stringency. The ability of SAMI and concurrent software to detect intragenic splicing aberrations and gene fusions was assessed, both on real data from a commercial control sample and simulated data generated with ASimulatoR. Nextflow pipeline and Singularity container recipe freely available under GPL 3 licence at https://github.com/HCL-HUBL/SAMI.