Identification of potential biomarkers and their correlation with immune infiltration cells in schizophrenia using combinative bioinformatics strategy

• Published in: Psychiatry research Vol. 314; p. 114658
• Main Authors: Li, Zhijun, Li, Xinwei, Jin, Mengdi, Liu, Yang, He, Yang, Jia, Ningning, Cui, Xingyao, Liu, Yane, Hu, Guoyan, Yu, Qiong
• Format: Journal Article
• Language: English
• Published: Ireland Elsevier B.V 01.08.2022
• Subjects: Bioinformatics analysis; Biomarkers; Gene expression; Microarray; Schizophrenia; Bioinformatics analysis; Schizophrenia; Biomarkers; Microarray; Gene expression
• ISSN: 0165-1781; 1872-7123
• DOI: 10.1016/j.psychres.2022.114658

•Our study is the first to use a combination of RRA, WGCNA and CIBERSORT algorithms to explore novel biomarkers associated with SCZ.•We presented gene expression changes associated with schizophrenia consistent across up to 10 independent cohorts of subjects, and revealed a significant correlation between key genes expression levels and the immune cells.•Moreover, the risk prediction model using those potential biomarkers achieved a high accuracy for diagnosis of SCZ. Many studies have identified changes in gene expression in brains of schizophrenia patients and their altered molecular processes, but the findings in different datasets were inconsistent and diverse. Here we performed the most comprehensive analysis of gene expression patterns to explore the underlying mechanisms and the potential biomarkers for early diagnosis in schizophrenia. We focused on 10 gene expression datasets in post-mortem human brain samples of schizophrenia downloaded from gene expression omnibus (GEO) database using the integrated bioinformatics analyses including robust rank aggregation (RRA) algorithm, Weighted gene co-expression network analysis (WGCNA) and CIBERSORT. Machine learning algorithm was used to construct the risk prediction model for early diagnosis of schizophrenia. We identified 15 key genes (SLC1A3, AQP4, GJA1, ALDH1L1, SOX9, SLC4A4, EGR1, NOTCH2, PVALB, ID4, ABCG2, METTL7A, ARC, F3 and EMX2) in schizophrenia by performing multiple bioinformatics analysis algorithms. Moreover, the interesting part of the study is that there is a correlation between the expression of hub genes and the immune infiltrating cells estimated by CIBERSORT. Besides, the risk prediction model was constructed by using both these genes and the immune cells with a high accuracy of 0.83 in the training set, and achieved a high AUC of 0.77 for the test set. Our study identified several potential biomarkers for diagnosis of SCZ based on multiple bioinformatics algorithms, and the constructed risk prediction model using these biomarkers achieved high accuracy. The results provide evidence for an improved understanding of the molecular mechanism of schizophrenia.