By the summer, the performance of the supercomputer at the SKOLKOVO Institute of Science and Technology is expected to rise to 1.5 petaflops.
Maxim Fedorov, the Director of Skoltech Center for Computational and Data-Intensive Science and Engineering, said that, currently, Skoltech has several hybrid clusters (GPU/CPU) with a total performance of about 100 TF. The upgrade will increase the peak performance of Skoltech’s supercomputer infrastructure by an order of magnitude, and this will open some qualitatively new opportunities. For example, when it comes to medicinal products, the supercomputer will enable the processing of databases with tens of billions of records describing the parameters of drug candidate molecules, which currently cannot be provided fast enough.
As for the pharmaceutical research in general, we should bear in mind that bioactive substances are inherently toxic.
“Poison is in everything, and no thing is without poison. The dosage makes it either a poison or a remedy.”
This maxim was expressed by Paracelsus. The search for new drug may not be safe. Therefore, it is very important to predict the toxicity of the new compound even before it is synthesized. This is the kind of research that is conducted by the scientific team of Professor Maxim Fedorov.
“By combining supercomputing technology, the methods of mathematical modeling and machine learning, we are trying to predict not only the bioactivity but also the toxicity of compounds,” he said. “This knowledge enables us to use a different degree of protection when synthesizing a new compound; for example, we can synthesize initially not an entire gram, but a much smaller dose, a nanogram or a milligram.”