Deep Ectrophrm AI system discovered 464 species of Escherichia coli enzymes (Shutterstock)
A joint Korean-American research team has successfully employed artificial intelligence to uncover the enzymatic secrets of one of the most important organisms, a bacteria that inhabit the intestines called Escherichia coli.
These bacteria were first discovered in 1885 by German pediatrician and bacteriologist Theodor Echirich, but research on them gained momentum in the early and until mid-twentieth century, due to their importance in scientific research and microbiology.
The intensive study began in the forties when it became a model organism for genetic research and molecular biology, after scientists found that it grows easily in laboratory conditions, has a rapid reproduction rate, and shares many genetic similarities with other organisms, including humans.
Despite this long history of studies, scientists have not yet understood what 30% of their proteins do, but a new artificial intelligence system called "Deep Ektransofrum" may help after researchers from the Korea Advanced Institute of Science and Technology and the University of California succeeded in using it to discover the functions of 464 unknown enzymes in these bacteria, the achievement was announced in the journal "Nature Communication".
Enzyme detection helps in employing them for the analysis of plastics or compound biosynthesis (Shutterstock)
What did previous studies do?
Enzymes are biological molecules that act as catalysts in living organisms, and their primary function is to accelerate and regulate chemical reactions necessary for life, and knowing what each of them does helps us understand how organisms work.
Many previous studies have used AI to discover and predict their functions, leveraging the capabilities of machine learning and generalized learning techniques. However, despite their success in predicting jobs, these studies have not been able to explain the reasons for this prediction, which greatly hindered the use of their results.
- Deep Ik System: Researchers have developed the deep-learning Deep Ek method to predict enzymatic classification numbers based on protein sequences. Enzymatic classification is a system developed by the International Union of Biochemistry and Molecular Biology to classify and regulate enzymes based on the specific chemical reactions that catalyze them.
- Machine learning algorithms: Various studies have used machine learning algorithms to predict enzyme functions, and these methods analyze sequence data, protein structures, and other biological features to predict the functions of enzymes, and these studies have reached promising results in accurately predicting enzyme activities and catalytic functions.
- Enzyme design: Applied AI to enzyme design to engineer function-specific enzymes for industrial or biomedical purposes using computational methods and machine learning algorithms.
- Functional genomics: AI-based approaches to functional genomics included large-scale prediction and characterization of enzymes, and these studies contributed to understanding metabolic pathways, identifying potential drug targets, and discovering enzymes relevant to biotechnology applications.
While these previous studies have been successful in predicting some of the functions of enzymes, they have not been able to explain why this prediction is made, which is why enzymes like black boxes are unknown. But the Deep Ectoscope system was able to understand the language of proteins, which helped overcome this problem.
During the study, the researchers announced that their new system could predict what enzymes would do based on their protein sequences, enabling them to quickly and accurately determine the functions of enzymes and explain why the prediction was.
The deep ectochoferm system can predict what coliform enzymes do based on their protein sequence (Shutterstock).
How does Deep Ectrophrm work?
In an exclusive email interview with "Al Jazeera Net", a professor at the National Research Laboratory for Metabolic and Molecular Bioengineering at the Korea Advanced Institute of Science and Technology and the lead researcher of the study, Sang Yup Lee, says: The "Deep Ektransofrum" tool works using the smart grid and the search system, if the smart grid cannot know what the protein does, the search system searches for similar proteins and guesses based on what those similar proteins do, which makes their new system good at guessing the functions that They carry out proteins and determine the cause of this guess, compared to other similar systems.
"It's a smart system for being able to explain why he's making some guesses, such as pointing out certain parts of proteins that help him make decisions, and this understanding that similar systems don't provide is very useful for making new things, such as eco-friendly chemicals, using these newly discovered proteins."
The Dip Ektranesophrm system was able to understand the language of proteins, which helped to understand why certain enzyme functions were predicted (Shutterstock)
Results valid for generalization
Although the study that led to the Deep Ektranesophrm tool was on Escherichia coli, it could be employed to work with the proteins of many organisms, which opens the door to many researches that reveal more functions of enzymes of organisms, and this means that we are facing a promising tool that paves the way for many practical applications.
"Enzymes like tiny machines in organisms that can break down plastics or biosynthesize various beneficial compounds, and the AI tool helps us find them faster and better by sorting out a lot of genetic information, finding more of it, which in turn leads to the creation of more environmentally friendly factories and the discovery of new things in biology," says Sang-Yop Lee.
The researchers are confident in the accuracy of their system, and he has done well at guessing the functions of proteins, even for those that are very different from those he was trained for, and this shows that he is good at detecting protein functions in general.
Source : Al Jazeera + Websites