Efficient evidence-based genome annotation with EviAnn.

Eukaryotic annotation system building exon-intron structures directly from transcript alignments or protein homology, rather than ab initio predictions

Machine-learning ab initio annotators long remained central due to scarce expression data. The authors develop EviAnn (Evidence-based Annotator), a strongly data-driven eukaryotic annotation system building exon-intron structures of coding and non-coding genes directly from transcript alignments or protein homology. On identical input, EviAnn outperforms references such as BRAKER3, MAKER2 and FINDER while using far less compute — a mammalian genome is annotated in under an hour on a multicore server. The tool is open source.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

A useful incremental improvement in genome annotation, with a clear speed and accuracy gain over references and open-source availability. Its clinical impact is indirect (quality of annotation references), but better annotation underpins all downstream interpretation.

Analysis by Dr Thibaut Benquey