How to spice up your bioinformatics skill set with AI
Published in Artificial Intelligence.
Incorporating machine-learning tools into data analysis can accelerate discovery and free up valuable time.
Artificial-intelligence programs can speed up monotonous tasks in research — and the learning curve is not too steep.
Image-analysis tools can do amazing things. Yet despite their power, Fernanda Garcia Fossa was frustrated. A biology PhD student at the State University of Campinas, Brazil, Garcia Fossa specializes in nanotoxicology. Image-based profiling of human cells is a core part of her research. But when she started out, the process was slow and error-prone.
“I spent a lot of my time analysing my images individually by hand, looking for differences and patterns,” Garcia Fossa explains. She was looking for evidence of the subtle effects of silver nanoparticles on liver cells. But the number of hours it took to compare scanned images of each cell one by one was overwhelming, she says. “I thought, there has to be a faster way to do this.”
Trawling online biology forums, she stumbled on CellProfiler, an image-analysis tool based on artificial intelligence (AI) developed at the Broad Institute of MIT and Harvard in Cambridge, Massachusetts. Within hours, she had identified an algorithm tailored to her needs, which she used to analyse her images automatically. “It was exciting,” she says. “Suddenly, I found I had more time to do other tasks related to my research, because the program was analysing all my images for me.”