Identification of Transcriptome Biomarkers for Severe COVID-19 with Machine Learning Methods.
Biomolecules
; 12(12)2022 11 23.
Article
in English
| MEDLINE | ID: covidwho-2123515
ABSTRACT
The rapid spread of COVID-19 has become a major concern for people's lives and health all around the world. COVID-19 patients in various phases and severity require individualized treatment given that different patients may develop different symptoms. We employed machine learning methods to discover biomarkers that may accurately classify COVID-19 in various disease states and severities in this study. The blood gene expression profiles from 50 COVID-19 patients without intensive care, 50 COVID-19 patients with intensive care, 10 non-COVID-19 individuals without intensive care, and 16 non-COVID-19 individuals with intensive care were analyzed. Boruta was first used to remove irrelevant gene features in the expression profiles, and then, the minimum redundancy maximum relevance was applied to sort the remaining features. The generated feature-ranked list was fed into the incremental feature selection method to discover the essential genes and build powerful classifiers. The molecular mechanism of some biomarker genes was addressed using recent studies, and biological functions enriched by essential genes were examined. Our findings imply that genes including UBE2C, PCLAF, CDK1, CCNB1, MND1, APOBEC3G, TRAF3IP3, CD48, and GZMA play key roles in defining the different states and severity of COVID-19. Thus, a new point of reference is provided for understanding the disease's etiology and facilitating a precise therapy.
Keywords
Full text:
Available
Collection:
International databases
Database:
MEDLINE
Main subject:
Transcriptome
/
COVID-19
Type of study:
Diagnostic study
/
Etiology study
/
Prognostic study
Limits:
Humans
Language:
English
Year:
2022
Document Type:
Article
Affiliation country:
Biom12121735
Similar
MEDLINE
...
LILACS
LIS