转录组
疾病
选择(遗传算法)
肺结核
表达式(计算机科学)
计算生物学
生物
医学
生物信息学
基因表达
计算机科学
内科学
人工智能
遗传学
病理
基因
程序设计语言
作者
Li Li,Ting Wang,J Q Liang,Hui Ding
出处
期刊:PubMed
日期:2025-07-12
卷期号:48 (7): 649-655
标识
DOI:10.3760/cma.j.cn112147-20250219-00092
摘要
Objective: To use high-throughput transcriptome sequencing data to screen for a set of non-invasive diagnostic biomarkers for assessing the risk of pulmonary tuberculosis (PTB) and predicting disease progression stages. Methods: A total of 37 effective microarray transcriptome datasets were collected from the National Center for Biotechnology Information (NCBI) from January 2015 to December 2024, covering five groups: control group, PTB group, extrapulmonary TB (EPTB) group, latent TB infection (LTBI) group, and other diseases (OD) group. Background correction and quantile normalization were performed using the robust multi-array average (RMA) method. Differential expression analysis was conducted using the "limma" package in R language, and protein-protein interaction analysis was performed using the STRING database. Additionally, a random forest model was trained to predict different stages of TB progression, and a biomarker signature for predicting PTB progression was constructed using the "ssgsea" method in the "GSVA" package. Results: Differential expression analysis revealed gene expression differences between the PTB group and other groups, resulting in a set of genes with PTB-specific expression, including C1QA(Complement C1 subunit A), IRF7(Interferon Modulator 7), LGALS3BP(galectin 3-binding protein), SERPING1(Complement C1 inhibitors), and UBE2L6(ubiquitin ligase E2 L6). These genes exhibited rich interaction relationships and were closely associated with biological processes and pathways related to PTB. Using the random forest model, these five specific genes were identified as potential biomarkers for predicting PTB patient progression. In the test set, the model showed good diagnostic performance at different stages, with an area under the receiver operating characteristic curve (AUC)>0.8, especially performing best at 5 months (AUC=0.992). Conclusion: Using transcriptome data, a classification model was established to assess PTB risk and predict disease progression stages, and five important biomarkers were identified that provide crucial references for the early diagnosis and treatment of PTB.
科研通智能强力驱动
Strongly Powered by AbleSci AI