Applications are mostly in cell biology where one has to identify the morphology of each protein molecule.

Applications are mostly in cell biology where one has to identify the morphology of each protein molecule. The morphology is directly related to the binding ability of the proteins. So when you develop a drug for treatment of a particular disease, if it should be a targeted medicine that binds only to the damaged cells - like in cancer cells for example, you need to know the morphology of the protein molecule. Now the protein is having a 3D structure. Because it is so small - in nano sizes - one can't pick up one protein and turn around to see its structure. The only way is to photograph them with electron microscopes. This was a breakthrough research and Jacques Dubochet, Joachim Frank and Richard Henderson were awarded the Nobel Prize in Chemistry 2017 for the discovery of cryo-electron microscopy. However, there was a problem. Proteins are not the only nanoparticles in the cryoEM image. Researchers had to be trained to pick up protein from the images by physically labelling what are proteins and what are not. This used to take weeks even for an expert. During one of our initial discussions, the National Center for Cell Sciences in Pune (NCCS) told us that this is the next hurdle in the automation of protein 3D reconstruction. There was no solution for it at that time. It was a long two years of hard work, However, a few months before we submitted our work 2 other groups came up with partial solutions to this issue. They needed some hand picked samples to automate the process. Our method was very different. We developed a method that could identify the transmittance of protein cells in general as compared to ice, dust, carbon, graphite etc that come as contaminants in the image. Thus, our method could identify each pixel that belongs to a protein with some confidence and label them without human touch! It took two rounds of thorough review by the reviewers to confirm that our method works and can be used even to discover the structure of new proteins. So, this work, CASSPER tool which is open source and free, is able to do the handpicking work that took weeks to a fraction of a second and with even more reliability than the human. The method will help scientists to fly on a rocket instead of their old bullock cart rides on protein morphology determination. That is what we did in a nutshell.