The Russian research institute AIRI has introduced an innovative tool, GENATATOR — a neural network capable of mapping genes with high accuracy based on DNA sequences. This is not just another algorithm, but a full-fledged AI system that automatically finds gene boundaries, determines the transcript type, and reconstructs the internal structure of the genome.
GENATATOR consists of several models that work in stages. First, the system searches for probable start and end points of transcripts, then analyzes the regions between them, refines the structure, and filters out questionable hypotheses. This approach significantly improves accuracy compared to classical methods that rely on searching for individual markers, such as start and stop codons or splicing signals.
The key advantage of GENATATOR is its versatility. Unlike traditional tools, it is not tied to searching for specific markers but instead memorizes and learns patterns in entire DNA sequences. This makes it applicable to genomic assemblies of non-model organisms for which detailed annotations are unavailable. The models were trained on human genes and those of 38 mammalian species, including walruses and elephants, but tests on fruit flies, thale cress, and baker's yeast showed excellent results.
Application and Availability
The tool is designed to identify protein-coding genes and long non-coding RNA genes. Its potential applications are vast: from studying evolution and discovering new genes to analyzing ancient animals. In medicine, GENATATOR could become indispensable for disease prediction, developing personalized drugs, and biotechnological projects.
The models have already been released as open access on the Hugging Face platform, along with a launched web service and an open leaderboard for quality assessment. Currently, the local version only works on CUDA GPUs with the float32 format, but this is a technical limitation that will likely be addressed in upcoming updates.
Expert commentary: GENATATOR represents a significant step forward in bioinformatics. The ability to accurately map genes without relying on annotations opens new horizons for research in genetics and medicine. It is especially important that the tool is available to the community — this will accelerate the development of personalized therapy and fundamental discoveries.