Efficient evidence-based genome annotation with EviAnn

Machine-learning-based ab initio gene finders have long been central to eukaryotic genome annotation, largely because gene expression data were historically expensive and limited. Modern sequencing technologies have transformed this landscape, making transcript evidence abundant and reliable, while proteins from many well-annotated genomes provide additional rich sources of evidence. Here we address the underutilization of these data by existing annotation tools by developing EviAnn (Evidence-based Annotator), an evidence-based eukaryotic gene annotation system. EviAnn takes a strongly data-driven approach, building exon-intron structures of protein coding and noncoding genes directly from transcript alignments or protein homology rather than relying primarily on ab initio predictions. Using identical input data, EviAnn consistently outperforms leading packages such as BRAKER3, MAKER2 and FINDER, while utilizing considerably less computer time. A mammalian genome can be annotated in under an hour on a single multicore server. EviAnn is open-source software available via GitHub at https://github.com/alekseyzimin/EviAnn_release and through Bioconda as 'eviann'.