SQANTI-evidence: isoform-resolved, evidence-tiered structural annotation of unannotated genomes from long-read RNA-seq

SQANTI-evidence annotates unannotated genomes from long-read RNA-seq, with its curated layer most precise in five of seven species and recovering twice as many exact isoforms as BRAKER in mammals.

• Why it matters: Long-read RNA captures full-length transcripts, but gene prediction pipelines often mis-curate long-read evidence, limiting accurate structural annotation when no prior resources exist.
• What they did: The workflow uses SQANTI3-curated transcripts as observed evidence (including isoforms and untranslated regions) and runs AUGUSTUS only at loci lacking read support.
• The result: In benchmarks across seven species, the curated layer was most precise in five, recovered 2× more exact isoforms than BRAKER in mammals, and placed 3' ends nearest the poly(A) site; it also annotated 13 ERGA datasets.

AI summary of the abstract · may contain errors · not medical advice · read the paper.