A Three-gene-based Type 1 Diabetes Diagnostic Signature

Area of Science:

• Genomics and Bioinformatics
• Immunology
• Computational Biology

Background:

• Type 1 diabetes is an autoimmune disease characterized by pancreatic beta-cell destruction and insulin deficiency.
• Current treatments require lifelong insulin injections, highlighting the need for improved diagnostic and management strategies.
• Identifying reliable biomarkers is crucial for early diagnosis and personalized treatment of type 1 diabetes.

Purpose of the Study:

• To identify potential diagnostic biomarkers for type 1 diabetes using bioinformatics analysis of gene expression data.
• To develop and validate a predictive model for type 1 diabetes diagnosis based on identified biomarkers.
• To explore the utility of machine learning in analyzing large datasets for disease biomarker discovery.

Main Methods:

• Collected and analyzed mRNA expression datasets (GSE50098, GSE9006) from peripheral blood samples of type 1 diabetes patients.
• Screened differentially expressed genes (DEGs) and performed Gene Ontology (GO) and KEGG pathway enrichment analysis.
• Constructed a protein-protein interaction (PPI) network to identify hub genes and developed a logistic regression model for sample classification.

Main Results:

• Identified 10 overlapping DEGs between the two datasets, with 7 hub genes identified through PPI network analysis.
• Developed a 3-gene logistic regression model (LTF, CAMP, PGLYRP1) with high diagnostic accuracy.
• Achieved an Area Under the Curve (AUC) of 0.8452 for the training set and 0.8083 for the testing set, demonstrating model efficacy.

Conclusions:

• Integrated bioinformatic analysis successfully identified key genes associated with type 1 diabetes.
• The developed logistic regression model based on LTF, CAMP, and PGLYRP1 shows promise as a diagnostic tool for type 1 diabetes.
• This study provides a foundation for developing novel, non-invasive diagnostic methods for type 1 diabetes.