A new AI toolkit – developed by scientists at the Francis Crick Institute and UCL – can explore how pathogens bring our cells with the precision of a trained scientist. The tool, called HRMan ("Herman"), is a readily accessible, open source code that can help to adapt various pathogens, including Salmonella enterica.
The toolkit utilizes deep neural networks to explore complex patterns in the images of pathogens and human cell interactions, drawing out the same detailed features that scientists make by hand.
Eva Frickel, head of the Krikka group who runs the project, said: "What was biometric before, it takes a lot of time for biologists, which takes us a matter of minutes on a computer, which allows us to learn more about infectious pathogens and how our bodies react to them faster and more accurately. HRMAn can actually see the interactions of host-pathogens as a biologist, but unlike us, he does not get tired and needs to sleep! "
Arthur Yakimovich, a researcher at Jason Mercer's Laboratory at MRC LMCB at UCL and the first author of the study, said: "Previous attempts to automate host-pathogen image analysis failed to capture this level of detail. By using the same types of algorithms in which self-driving cars are running, we have created a platform that increases the accuracy of high volume biological data analysis that revolutionizes what we can do in the lab. AI algorithms are useful when the platform evaluates image-based data in a way that would train a specialist. It's also very easy to use, even for scientists who have barely any knowledge of coding. "
Daniel Fish, Crick Ph.D. A student and first author of the study said: "Our team uses HRMAn to answer specific questions about host-pathogen interactions, but there are significant out-of-field effects, HRMAn can analyze any fluorescence image that makes it suitable for batches of different areas of biology, including cancer research. "
The study is published in the eLife Open Access Magazine, which includes a link to download the platform and access to tutorial videos.