With the help of artificial intelligence, scientists have studied millions of different compounds and identified many that could act as drugs that destroy SARS-Cov2. These tests can also help you find cures for other diseases.
Artificial intelligence is entering medicine more and more boldly. It begins to use it, among others to search for new drugs, which may help in the fight against the current epidemic. An example is the work of a team from the University of California, Riverside described in the pages of the magazine “Heliyon”.
«There is an urgent need to find effective measures to treat or prevent COVID-19. We have developed a pipeline processing system for drug discovery, which found a number of potential substances »- says Prof. Anandasankar Ray.
This system is based on artificial intelligence that learns to predict the action of various compounds. “Pipeline systems for analyzing drugs, such as ours, are extremely important as the first step in a systematic search” – adds the expert.
The researchers explain that the priority is the use of drugs already approved by the FDA and used in other diseases against COVID-19. However, there is also a great need for new drugs to inhibit viral entry and multiplication. In both cases, a new method, an algorithm that searches for substances that interact with some of the 65 human proteins relevant to the body’s interaction with the virus, can help in both cases. Algorithms do this by analyzing the three-dimensional structure of the compounds under study.
“These 65 proteins are rather heterogeneous and they are important in many other diseases, including cancer. In addition to reusing the existing drugs that affect these proteins, we also wanted to find new substances that are currently not well researched “- emphasizes the creator of the system Joel Kowalewski.
Using the new method, scientists analyzed more than 10 million chemical molecules on the market, described in a database containing 200 million substances. Thus, they found the best interaction with 65 proteins important for the development of infection.
They also examined 100. substances already approved by the FDA (as drugs or food additives). Using computer models, they also pre-analyzed the potential toxicity of the tested molecules, which led to the rejection of harmful compounds.
Ultimately, they identified the most promising substances, some of which interacted simultaneously with even two or more proteins, important in the event of SARS-Cov2 infection.
“The compounds that excite me the most are volatile, which means that it is possible to develop inhaled drugs” – emphasizes Prof. Ray.
Scientists point out that their method is much cheaper and faster than typical biological tests on cells. “Our database can allow the rapid identification and testing of new and safe treatment strategies for COVI-19 and other diseases for which these 65 proteins are important. Although we were motivated by the COVID-19 pandemic, we expect that our predictions of 10 million molecules will accelerate the search for drugs against other diseases as well, ”says Kowalewski.
- Compulsory and free COVID-19 vaccine? «This is the most convenient solution»
“Historically, disease management has become more and more complex over time as we gain a better understanding of the diseases and the genetic predisposition contributing to their development and severity of symptoms. The machine learning approach can play a large role in anticipating therapeutic possibilities by providing researchers with additional options for further study. While this approach relies heavily on experimental data, virtual screening can help researchers ask new questions and discover new information, ‘adds the researcher.
Now scientists are applying for funding and looking for collaborators to conduct further studies discovered in cells, animals, and ultimately humans.
Read also:
- There will be a vaccine, but will people want to get vaccinated?
- Iran: “We are already testing three to four vaccines in humans”
- Tuberculosis vaccine will protect against COVID-19?