# Machine learning-driven discovery of NETs-associated diagnostic biomarkers and molecular subtypes in tuberculosis

## TL;DR

This study uses machine learning to identify key genes linked to NETs in tuberculosis, revealing potential biomarkers and molecular subtypes for better diagnosis and treatment.

## Contribution

The study introduces a novel machine learning approach to identify core NETs genes and molecular subtypes in tuberculosis.

## Key findings

- Six core NETs genes were identified using an ensemble of 113 machine learning algorithms.
- Patients were stratified into two subtypes with distinct immune infiltration profiles.
- RT-qPCR validated the differential expression of key NETs core genes.

## Abstract

NETs constitute a pivotal mechanism in the pathogenesis and progression of TB. Despite their recognized importance, the genetic underpinnings of NETs in TB remain inadequately elucidated. Accordingly, the present study endeavors to delineate the molecular characteristics of NRGs in TB, with the objective of reliably identifying associated molecular clusters and biomarkers.

Gene expression profiles were analyzed from integrated datasets retrieved from the GEO database. Differential analysis, WGCNA, and an ensemble of 113 machine learning algorithms were employed to identify the core NETs genes. Subsequently, TB patients were stratified into distinct subtypes based on the expression profiles of these core genes, and the differences in immune infiltration characteristics among the subtypes were systematically compared. Finally, RT-qPCR was utilized to validate the differential expression of the key NETs core genes.

Analysis of the integrated GSE83456 and GSE54992 datasets yielded 630 DEGs. WGCNA subsequently identified a module comprising 1,252 genes, from which 26 key NETs genes were extracted via intersection with known NRGs. Among the ensemble of 113 machine learning methods, the “StepgIm[both]+RF” algorithm demonstrated superior performance, ultimately identifying six core NETs genes. Consensus clustering based on the expression profiles of these core genes stratified patients into two distinct subtypes. Functional enrichment analysis further underscored the predominance of immune-related pathways in subtype B. Moreover, immune infiltration analysis revealed marked differences in immune cell composition between the subtypes, thereby confirming a close association between the core NETs genes and these immunological disparities.

Core NETs genes are pivotal in the pathogenesis and progression of tuberculosis, and they hold significant promise as novel biomarkers for the early diagnosis and targeted treatment of TB.