Ó³»­´«Ã½

MOTIVATION: Today, we know the function of only a small fraction of all known protein sequences identified. This problem is even more salient in bacteria as human-centric studies are prioritized in the field and there is much to uncover in the bacterial genetic repertoire. Conventional approaches to bacterial gene annotation are especially inadequate for annotating previously unseen proteins in novel species since there are no proteins with similar sequence in the existing databases. Thus, we need alternative representations of proteins. Recently, there has been an uptick in interest in adopting natural language processing methods to solve challenging bioinformatics tasks; in particular using transformer-based language models to represent proteins has proven successful in tackling various challenges. However, there are still limited applications of such representations in bacteria.RESULTS: We developed SAP, a novel synteny-aware gene function prediction tool based on protein embeddings, to annotate bacterial species. SAP distinguishes itself from existing methods for annotating bacteria in two ways: (i) it uses embedding vectors extracted from state-of-the-art protein language models and (ii) it incorporates conserved synteny across the entire bacterial kingdom using a novel operon-based approach proposed in our work. SAP outperformed conventional annotation methods on a range of representative bacteria, for various gene prediction tasks including distant homolog detection where the sequence similarity between training and test proteins was 40% at its lowest. SAP also achieved annotation coverage on par with conventional structure-based predictors in a real-life application on genes of unknown function.