Game theory vs. "resistance genes"
Machine learning helps find drug-resistant bacteria
Tatiana Matveeva, "Scientific Russia"
The new software (Prediction of Antimicrobial Resistance via Game Theory, PARGT), developed by scientists from Washington State University (USA), helps to identify the genes that make bacteria resistant to drugs, the press service of the university reports. The results of the work are published in the journal Scientific Reports (Chowdhury et al., PARGT: a software tool for predicting antimicrobial resistance in bacteria).
Antimicrobial resistance (AMR) occurs when bacteria or other microorganisms develop mutations as a result or acquire genes that encode drug resistance mechanisms. For example, bacteria that cause staphylococcal or streptococcal infections or diseases such as tuberculosis and pneumonia have developed strains that are not affected by medications, including antibiotics.
As large-scale genetic sequencing has become easier today, researchers are looking for genes responsible for drug resistance. In order to more accurately and faster identify AMR-resistant genes, scientists have developed a special machine learning algorithm.
The program uses signs of "resistance proteins" rather than similarity of gene sequences to identify AMR genes. The researchers turned to game theory, a tool that is used in several fields (especially economics) to simulate strategic interactions between players. Oddly enough, game theory helps identify AMR genes. Using a machine learning algorithm and a game theory approach, the researchers studied the interaction of several characteristics of genetic material, including its structure and the physicochemical and compositional properties of protein sequences, and not just the similarity of sequences.
"The advantage of this program is that we can actually detect AMR in recently sequenced genomes," the authors note.
The Washington State University team examined the "resistance genes" found in Clostridium, Enterococcus, Staphylococcus, Streptococcus and Listeria species. These bacteria are the cause of many major infections and infectious diseases, including staphylococcal infections, food poisoning, pneumonia and life-threatening colitis. They were able to accurately classify the "resistance genes" with an accuracy of up to 90%.
They have developed a software package that other researchers can download and use to search for AMR in big data of genetic material. The software can also be improved over time. The algorithm can also be retrained when new data and sequences appear.
Portal "Eternal youth" http://vechnayamolodost.ru