Identification of secretory protein related biomarkers for primary biliary cholangitis based on machine learning and experimental validation

TL;DR

This study identifies secretory proteins as potential biomarkers for early diagnosis of primary biliary cholangitis using machine learning and experimental validation.

Contribution

The study introduces novel secretory protein biomarkers for PBC diagnosis, validated through machine learning and animal experiments.

Findings

Machine learning identified 14-18 key diagnostic genes with an AUC of 0.96.

Upregulated genes like CSF1R, PLCH2, and SLC38A1, and downregulated CST7 were highlighted as significant.

Animal experiments confirmed the bioinformatics findings.

Abstract

Primary biliary cholangitis (PBC) is a chronic autoimmune liver disease that causes bile duct damage, liver fibrosis, and cirrhosis, significantly affecting patients’ lives and healthcare costs. Early diagnosis is critical but is hindered by the limited sensitivity of existing biomarkers, particularly in patients who are negative for anti-mitochondrial antibodies. This limitation underscores the need for more reliable biomarkers. Our study focuses on secretory proteins as potential diagnostic biomarkers and aims to elucidate gene expression profiles associated with PBC using bioinformatics methods and machine learning. We identified 827 downregulated and 639 upregulated genes related to mitochondrial function and immune pathways. Additionally, Weighted Gene Co-expression Network Analysis revealed a blue module comprising 1,949 genes linked to PBC. Machine learning identified between 14 and 18 key diagnostic genes. Using a Gaussian Mixture Model, we achieved an area under the curve of 0.96, indicating excellent diagnostic performance. Notable genes included the upregulated CSF1R, PLCH2, and SLC38A1, as well as the downregulated CST7. Animal experiments further supported these bioinformatics findings. Our research highlights secretory proteins as promising biomarkers for the early diagnosis of PBC, with potential applications in developing precise diagnostic tools and personalized therapies. This work paves the way for future studies involving larger cohorts and multi-omics data.

Figures