Identification of Transcriptome Biomarkers for Severe COVID-19 with Machine Learning Methods

Area of Science:

• Genomics
• Computational Biology
• Infectious Disease Research

Background:

• The global spread of COVID-19 necessitates personalized treatment strategies due to varying patient symptoms and disease severity.
• Accurate classification of COVID-19 states and severity is crucial for effective patient management.

Purpose of the Study:

• To employ machine learning techniques for identifying gene expression biomarkers that can accurately classify COVID-19 across different disease states and severities.
• To uncover essential genes and their molecular mechanisms contributing to COVID-19 pathogenesis and progression.

Main Methods:

• Analysis of blood gene expression profiles from COVID-19 patients (with and without intensive care) and non-COVID-19 controls.
• Application of Boruta and Minimum Redundancy Maximum Relevance (mRMR) for feature selection to identify relevant genes.
• Utilizing incremental feature selection to discover essential genes and construct robust classification models.

Main Results:

• Identification of a set of key genes, including UBE2C, PCLAF, CDK1, CCNB1, MND1, APOBEC3G, TRAF3IP3, CD48, and GZMA, as significant biomarkers.
• Demonstration of these genes' roles in differentiating between various COVID-19 disease states and severity levels.
• Examination of the molecular mechanisms and biological functions associated with the identified essential genes.

Conclusions:

• The identified genes serve as crucial indicators for classifying COVID-19 severity and disease states.
• These findings provide a novel reference for understanding COVID-19 etiology.
• The study facilitates the development of precise therapeutic interventions for COVID-19 patients.