Analysis of isoform complexity in pan-transcriptome graphs with atroplex

Alternative splicing of precursor mRNA lets a single gene encode multiple isoforms by joining exons in different combinations. Long-read sequencing resolves this isoform diversity across tissues, cohorts, and conditions. However, the resulting pan-transcriptomes are structurally complex, and their analysis requires repeatedly searching the full catalogue, which is impractical without a queryable index. As splicing patterns differ across conditions, a structure is needed that captures the connectivity between exons, not just their coordinates, so isoforms can be compared by structure across cohorts.

We present atroplex, a framework that indexes pan-transcriptome annotations and transcript isoforms in a combined spatial index and graph overlay, capturing both exon coordinates and splice connectivity. atroplex classifies query transcripts against the index, tracks per-sample isoform presence, and enables crosscohort isoform comparison. We indexed 21,005 samples spanning multiple reference resources into a single queryable structure, yielding a comprehensive map of isoform complexity that supports improved transcript discovery and structural comparison across cohorts.

atroplex is licensed under GPLv3 and available at https://github.com/ylab-hi/atroplex.