Metabolomic Insights into Tuberculosis: Machine Learning Approaches for Biomarker Identification

Abstract

The lung parenchyma is largely impacted by the infectious condition known as pulmonary tuberculosis (pulmonary TB) when the immune system creates a wall around the germs in the lungs, a tiny, hard bulge known as a tubercle develops, earning the disease the name tuberculosis. Although the majority of TB germs target the lungs, they can also harm other bodily organs. The identification of TB biomarkers, which are crucial for diagnosis, treatment monitoring, risk analysis, and prognosis, has been the subject of extensive research. Differences in metabolites between normal cells and tuberculosis are considered to be able to support the diagnosis of tuberculosis. Metabolite data was taken from the Metabolomic workbench and further identification and prediction were carried out in silico. A total of 44 samples found 69 metabolites which were then carried out further analysis. Found as many as 5 metabolites that play an important role in tuberculosis. Of the 5 metabolites, 2 candidate biomarkers were found which are known to have potential as biomarkers. The candidate biomarkers for these metabolites are trans-3-methyluric acid and nicotinic acid. However, this simulation needs further testing to obtain more accurate biomarkers and support the diagnosis.