Bacteria often need to secrete stuff (e.g., proteins) 🦠. To do this, they frequently use large complexes known as secretion systems. 💉 These secretion systems are important, as they help bacteria accomplish a lot of tasks. For example, they allow bacteria to grow, survive, and adapt to changing environmental conditions, and they facilitate interactions with hosts (e.g., humans) and other organisms (including other bacteria). Among bacterial pathogens, secretion systems play crucial virulence-promoting roles (e.g., by exporting protein toxins and effectors that directly target host cells; by promoting attachment/adhesion; via nutrient acquisition; by enabling biofilm formation) and thus are key to understanding how bacterial pathogens cause disease. 🤒

In collaboration with Johan Henriksson’s group at Umeå University 🇸🇪, we developed Sismis, a scalable, interpretable, machine learning-based tool, which detects and classifies secretion systems in bacterial (meta)genomes with high accuracy and speed. 🦇 To showcase its speed and scalability, we used Sismis to mine nearly 700k prokaryotic (meta)genomes for secretion systems…the largest dataset of its kind! 🧬

Check out the preprint, try Sismis for yourself, or browse the Sismis Atlas, our interactive atlas with Sismis secretion systems from 700k (meta)genomes! 🦇