Multi-Perspective Natural Vector: A Novel Method for Viral Sequence Feature Extraction.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• The exponential growth of biological data necessitates advanced sequence analysis techniques.
• Traditional pairwise alignment methods are computationally expensive and inefficient for large datasets.
• Existing alignment-free methods like natural vector (NV) and k-mer primarily use a 1D sequence approach.

Purpose of the Study:

• To introduce a novel alignment-free (AF) method, Multi-perspective natural vector (MNV), that incorporates 3D structural features of DNA.
• To develop an efficient method for processing large-scale genomic data, including variant classification and clustering.
• To enhance the accuracy and efficiency of genomic sequence analysis.

Main Methods:

• Developed the Multi-perspective natural vector (MNV) method, mapping genome sequences into a unified geometric space based on 3D architectural features.
• Evaluated MNV's performance on diverse datasets of varying sizes and types.
• Compared MNV against established NV and k-mer methods in terms of data processing and classification accuracy.

Main Results:

• MNV achieved a 100% convex hull separation ratio across datasets, outperforming NV and k-mer methods in lower dimensions.
• MNV demonstrated superior classification accuracy, reaching 99.55% for SARS-CoV-2 and 98.78% for poliovirus datasets using neural networks.
• The method effectively handles large genomic data, showcasing computational efficiency.

Conclusions:

• MNV offers a significant advancement in alignment-free sequence analysis by integrating 3D structural information.
• The method provides a robust and efficient framework for large-scale genomic data analysis and classification, particularly for viral genomes.
• MNV presents a promising tool for future bioinformatics research and applications requiring high-throughput genomic analysis.