Accepted_test

Application of machine learning algorithms to study the gut microbiota of patients with depressive disorders

by Galanova Olesya O. | Kovtun Alexey S. | Vavilov Institute of General Genetics, Russion Academy of Sciences, 119991 Moscow, Russia Moscow Institute of Physics and Technology, State University, 141701 Dolgoprudny, Russia | Vavilov Institute of General Genetics, Russion Academy of Sciences, 119991 Moscow, Russia

Abstract ID: 780

Event: BGRS-abstracts

Sections: [Sym 8] Section “Biotechnology through the lens of the microbiome”

Major depressive disorder (MDD) is the most prevalent mental disorders worldwide. Factors causing the pathogenesis of MDD include gut microbiota (GM), which interacts with the host through the gut–brain axis. Here we analyzed whole metagenome sequencing data to assess changes in both the composition and functional profile of GM. We looked at the GM of 36 MDD patients, compared with that of 38 healthy volunteers. We determined taxonomic and functional changes in the metagenomic signatures of MDD patients and healthy volunteers using the random forest algorithm. Decreased abundance was observed in the patients’ group for the genes encoding asparagine synthetase asnA, glutamate synthase gltB and gltD and spermidine synthase in species Feacalibacterium prausnitzii. Application of three machine learning algorithms (random forest, elastic network, YOLO) allowed for creation of the classifier, which can be used for depression diagnostics of patients by the state of their GM according to the metagenomic signature. The best results were observed for the YOLO algorithm with the average accuracy higher than 90%.