New AI tool speeds up gene annotation across many species
The study released by CEPLAS scientists Alisandra Denton (former member), Sebastian Triesch, Niklas Kiel, Nima Saadat, Oliver Ebenhöh, Björn Usadel Andreas Weber and others introduces Helixer, a novel deep-learning plus hidden-Markov-model–based tool for ab-initio gene prediction in eukaryotic genomes.
Helixer can produce highly accurate gene models from raw DNA sequences — without needing additional data like RNA-seq or homology — making it applicable for fungi, plants, vertebrates, invertebrates and more.
Benchmarking shows that Helixer’s predictions match or even surpass existing annotation tools and closely align with expert-curated references, enabling rapid, reliable genome annotation immediately after sequencing.