A Computational Framework for Pattern Detection on Unaligned Sequences: An Application on SARS-CoV-2 Data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• The increasing volume of genome sequences necessitates advanced pattern detection methods for evolutionary analysis.
• Traditional phylogenetic reconstruction relies on multiple sequence alignment and positional homology, which can be challenging with large datasets.
• Delineating species and extracting evolutionary signals from vast biological data remains a complex computational problem.

Purpose of the Study:

• To propose a novel computational framework for identifying meaningful features in unaligned sequence data.
• To leverage unsupervised learning techniques for characteristic k-mer discovery across various k-values.
• To utilize identified k-mers for sequence clustering and analyzing k-mer distribution shifts.

Main Methods:

• Development of a Python-based computational framework utilizing unsupervised learning.
• Identification of characteristic k-mers from unaligned sequence data for a range of k-values.
• Application of k-mers as features for clustering sequences and analyzing k-mer distribution patterns.

Main Results:

• Successful identification of characteristic k-mers within unaligned genome sequences.
• Effective clustering of sequence data based on identified k-mer features.
• Demonstration of analyzing changes in k-mer distributions across different k-values.

Conclusions:

• The proposed framework offers a promising approach for grouping sequence data based on inherent features.
• This method facilitates the study of evolutionary signals by analyzing k-mer distribution dynamics.
• The open-source Python software provides a freely accessible tool for bioinformatics research.