Related Experiment Video
Updated: May 9, 2025
Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis
Published on: June 18, 2020
Single Cell Transcriptomics Genomics Based on Machine Learning Algorithm: Constructing and Validating Neutrophil
Jia Yu1, Tiantian Xiao1, Yun Pan2
1Department of Internal Medicine, Dongguan Hospital of Integrated Chinese and Western Medicine, Dongguan, Guangdong Province, People's Republic of China.
Background:
Neutrophil trap (NET) is an important feature of chronic inflammatory diseases. At present, there are still few studies to explore the characteristics of NET in different chronic obstructive pulmonary disease (COPD) patients. This study aimed to identify NET signature genes in different COPD patients.
Methods:
We analyzed single-cell RNA sequencing data from COPD and non-COPD individuals to identify differentially expressed neutrophil genes. Machine learning algorithms were applied to construct models A and B, specific to smoking and non-smoking COPD patients, respectively.
Results:
Through single-cell cluster analysis, 165 neutrophil characteristic genes in COPD group were successfully identified. Model A, consisting of key genes CD63, RNASE2, ERAP2, and model B, consisting of GRIPAP1, NHS, EGFLAM, and GLUL, were validated internally and externally, showing significant risk scores and good diagnostic efficacy (AUC: 60.24-87.22). Alveolar lavage fluid in patients with COPD was studied and confirmed higher expression levels of RNASE2 and NHS in severe COPD patients.
Conclusion:
The study successfully developed NET signature gene models for identifying smoking and non-smoking COPD respectively, with validated specificity and predictive power, offering a foundation for personalized treatment strategies.
