The researchers at Chalmers University of Technology, Sweden recently came up with yet another promising way in which artificial intelligence can be deployed. The work published in the journal Nature Machine Intelligence throws light on an innovative method by which synthetic proteins can be generated using artificial intelligence. This is expected to be a promising breakthrough in the field of science that completely transforms the rubrics of health care.
Proteins-the large and complex molecules- holds a crucial place whether it be in the industrial domain or our daily lives. By naturally building, modifying and breaking down other molecules inside the cells, they play a key role in all living cells. There are a lot of drugs centered around proteins, one common and important example being insulin, the drug prescribed for diabetes. Proteins are major constituents of some very effective cancer medicines. And it shouldn’t come as a surprise that the antibody formulas used for the treatment of COVID-19 is also protein based.
The latest research that uses artificial intelligence to generate synthetic proteins, exhibits immense potential that can pave way for a very promising future characterized by efficiency, effectiveness and cost saving and faster drug development that ensures better health care.
Protein engineering, at present is largely dependent on random mutations that are introduced to protein sequences. However, this method has a downside. With the introduction of each mutation, there is considerable decline in the protein activity. This engineering process is extremely slow, characterized by experiments that are time consuming and costly with the additional chore of screening for millions of variants. This is where the speed of artificial intelligence and its accuracy fuses together to create a method that is rewarding and effective. With the help of artificial intelligence, the transformation from a computer design to an actual working protein can happen in the course of just a few weeks.
This innovative method was developed by Aleksej Zelzeniak,(assistant professor, Department of Biology and Biological Engineering, Chalmers University of Technology) and his team. This AI-centered approach which uses generative deep learning is called ProteinGAN.
A large amount of data derived from well studies proteins is given to the AI, which then attempts to study the data and create new proteins. Another part of the AI, simultaneously ensures if or not the synthetic proteins are fake by continuously sending the proteins back and forth in the system until it reaches a point where the AI can no longer distinguish between natural and synthetic proteins. This method enables the production of protein variants that are highly diverse with physical properties that edges naturalistic.
This novel approach will prove to be of immense help in the efficient development of industrial enzymes as well as that of protein based therapies like vaccines.
According to Martin Engqvist who also works at the Biology and Biological Engineering department, this new approach will help in developing a cost effective and environmentally sustainable industrial process. He also stresses the benefits and opportunities provided by a working environment that blends together computer science and biology.
In the due course, researchers will work on different methods to improve the properties of the proteins endowing them with better stability and efficiency, which will be of great benefit to proteins that are used in industrial technology.