Related Experiment Video
Updated: Jul 10, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.8K
Feature selection for effective prediction of SARS-COV-2 using machine learning.
1Nitte (Deemed to be University), Department of Molecular Genetics & Cancer, Nitte University Centre for Science Education & Research (NUCSER), Mangalore, Karnataka, India.
Genes & Genomics
|November 20, 2023
Summary
Classifying SARS-CoV-2 variants using machine learning (ML) aids early detection and transmission reduction. This study developed an ML model using proteomic data to predict disease severity and identify high-risk variants.
Area of Science:
- Computational Biology
- Machine Learning Applications
- Virology
Background:
- Emerging SARS-CoV-2 variants necessitate rapid classification for early detection and reduced transmission.
- Genomic and proteomic data integration with machine learning (ML) for SARS-CoV-2 classification remains underutilized.
Purpose of the Study:
- To develop a machine learning (ML) model for classifying SARS-CoV-2 strains based on proteomic evolutionary information.
- To generate a disease severity model utilizing nucleoprotein and amino acid charge/basicity distributions.
Main Methods:
- Utilized GISAID data for sequence and clinical information, incorporating proteomic calculations.
- Employed Select K-Best feature selection, cross-validated with testing sets, and utilized BIRCH clustering for strain analysis.
Main Results:
- Four out of six ML models achieved successful training and testing.
- Extra Trees algorithm yielded a micro-averaged F1-score of 74.2% and AUC-ROC of 73.7%.
- Feature selection improved ROC AUC to 76.4%, with an overall accuracy of 86.9%.
Conclusions:
- Identified unique features via ML for classifying SARS-CoV-2 disease severity.
- The developed ML approach shows potential for predicting risks associated with new SARS-CoV-2 variants.
