Machine learning-based identification and validation of amino acid metabolism related genes as novel biomarkers in chronic kidney disease.

Objectives: Chronic kidney disease (CKD) is a progressive illness with a high rate of morbidity and mortality with no proven therapy. Alterations of amino acid(AA) metabolism are associated with the incidence and progression of CKD. To characterize the potential value of AA metabolism related genes in the diagnosis and progression of CKD. Methods: We filtered the key genes associated with AA metabolism based on the least absolute shrinkage and selection operator (LASSO) and SVM algorithm. Then, we constructed logistic regression models and evaluated the accuracy and specificity by nomogram analysis and DCA. Also, we mapped the ROC curves.Meanwhile, in order to determine the underlying mechanism and relevant biological features of CKD, we conducted differential analysis between high and low risk subgroups in CKD. Moreover,we employed ssGSEA algorithm to evaluate the infiltration abundance of immune cells and calculated the correlation among the immune cells with the key genes. Finally,we validated the expression and clinical relevance of amino acid metabolism key genes via cultured cells and clinical data. A total of six key genes related to amino acid metabolism were identified, including ALDH18A1, CENPF, CSAD, CTH, CYP27B1, HBB. Results: All six genes exhibited promising diagnostic capabilities (AUC:0.7 to 0.9). Immune cells such as Activated CD4 Conclusions: Our study identified key genes involved in amino acid metabolism associated with immune cells infiltration and renal function in CKD, which may be potential biomarkers for the diagnosis and prognosis of CKD.