Proteins are fundamental to life on our Planet, and are also crucial in our wellbeing (and that goes as well for all life on Earth). Proteins interact in ways that are dependant on their "shape": contrary to what it would be normally expected their properties are not a consequence of the "stuff" (atoms) they are made of but solely of their shape (which in turns have to do with the atoms forming them although it is more than that).
Hence it is very important to understand their shape but this is extremely complex. A protein is made up of thousands of atoms and has a very convoluted structure. The structure in itself is extremely difficult to photograph because of the tiny dimensions involved (a whole protein is measured in nanometre, up to 100nm, and its individual components are a few tenth of a nanometer, in the Angstrom range), so the only way to see its structure is to simulate it using a computer but the computation requires huge amount of processing power, requiring hours of supercomputer processing.
A work around has been the creation of Rosetta@Home, a massive distributed computing systems leveraging on thousands of volunteers offering processing time of their computer. More recently the project Folding@Home harvest the processing power of smartphones to simulate the folding of proteins (see figure).
Yet, even with a supercomputer at hand, or leveraging the Rosetta virtual computer, the discovery of protein folding remains a challenge.
Now a team at the University of Toronto Scarborough has come up with an algorithm, based on deep learning structures, that can render the 3D structure of proteins in just over an hour using a good PC. The magic is due to the software. The protein is photographed over and over, thousands of times, using a cryo-Electronic Microscope that has the capacity of resolving at the nanometer scale. The photos are processed by the software and combined to create the 3D structure.
The researchers point out that this breakthrough should give a boost to our understanding of proteins, and hence of life itself. Besides, it should enable the design of new drugs and the experimentation of several alternatives at the "bit level" providing a fast track to find effective drugs (which in turns means lower cost and the possibility to fight rare diseases).
Another amazing result courtesy of ICT!