Compression-Complexity Measures for Analysis and Classification of Coronaviruses.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Rapid identification of viral pathogens is critical during global pandemics like COVID-19.
• Traditional sequence alignment methods are too slow for real-time pandemic analysis.
• Information theory and data compression offer powerful tools for analyzing biological sequences.

Purpose of the Study:

• To investigate compression-complexity measures (Effort-to-Compress and Lempel-Ziv complexity) for genomic sequence analysis.
• To apply these measures for the phylogenetic analysis of mammalian and viral genomic datasets.
• To develop an automated classification system for COVID-19-causing coronaviruses using machine learning.

Main Methods:

• Utilized compression-complexity based distance measures (ETC and LZ complexity) for sequence analysis.
• Reproduced phylogenetic trees for mammalian and various coronavirus datasets, including SARS-CoV-2.
• Employed machine learning classifiers (SVM, Linear Discriminant, KNN) for automated genome sequence classification.

Main Results:

• Successfully reproduced phylogenetic relationships in diverse genomic datasets.
• Achieved high classification accuracy (98%) for distinguishing COVID-19-causing coronaviruses from others.
• Demonstrated excellent sensitivity (95%) and specificity (99.8%) in classifying viral sequences.

Conclusions:

• Compression-complexity measures are effective for rapid genomic sequence analysis and phylogenetic reconstruction.
• The developed machine learning approach enables efficient and accurate automatic classification of coronaviruses.
• This methodology holds potential for real-time identification and characterization of viral strains and mutants in pandemic scenarios.