AI can Design Anti-biotic in a Few Days


The new AI (Artificial Intelligence) system of IBM can help in designing treatments for the COVID-19.

Any scientists or researcher wanting to find a specific treatment for a new disease needs a lot of time as there is hard work along with effort to be put in. You have thousands and thousands of molecules in your lab and you need to test which one works best for the treatment. You might find some that can help in killing the specific bacteria but you might not be able to use them as these might be toxic to human beings. This process is not only long but also expensive and aggravating.

IBM is using the ‘fast’ way to find treatments for different diseases by using the AI method. This AI system will be able to generate a specific design of the molecules for new antibiotics.

The researchers have already managed to design two antimicrobial peptides (small molecules that can kill the bacteria). Under normal circumstances this process would take years for the scientists. This was done in a few days by the AI system.


Antibiotic Resistance a Problem

New antibiotics can be life saving with great results. We depend on these for treating various diseases like tuberculosis, and other. These are also used for keeping you safe during surgeries and other procedures.

At present, experts have warned of a ‘post antibiotic’ era. This is when the existing antibiotics are getting useless. This crisis is due to the over-use of antibiotics in farm animals, crops and also humans.

Using excessive antibiotics, the bacteria can adapt to the drugs. They can then morph into superbugs that are anti-biotic resistant. These tend to render the drugs ineffective.

According to the latest report, this pandemic of the COVID-19 has only aggravated the problem. Doctors are inclined to prescribe antibiotic to the patients with the smallest of problems. Though it is a known fact that antibiotics cannot help viral infections, these antibiotics are prescribed against any secondary infection due to the COVID-19.

Working of the AI system of the IBM

This system of IBM works on a generative model. This is divided into 3 steps.

  1. The researchers use a massive database of peptide molecules that are known.
  2. The AI pulls out the information from this database and analyses the different patterns for figuring out the relationship between the said molecules and the properties. They find that a specific structure or even a composition of a molecule performs a specific function. This helps the system to ‘learn’ the molecule design rules.
  3. The researchers can then tell the AI what properties are needed from the new molecule. They also have the freedom to feed in constraints.

This does not mean that there will be magical solutions by the AI, but it can advance a computational strategy for solving problems that can yield the much needed benefits and also save lives.

The IBM researchers are hopeful that this use of AI can have broad applications.