Identity: rapid alignment-free prediction of sequence alignment identity scores using self-supervised general linear...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Pairwise global alignment is crucial for sequence analysis but computationally expensive for large datasets.
• Traditional alignment algorithms have quadratic time complexity, limiting their scalability.
• Many applications require only identity scores, not full alignment visualizations.

Purpose of the Study:

• To introduce Identity, a novel tool for calculating pairwise DNA sequence identity scores.
• To enable efficient analysis of large sequence datasets using alignment-free methods.
• To provide a faster and scalable alternative to existing sequence alignment tools.

Main Methods:

• Utilizes alignment-free methods and self-supervised general linear models.
• Predicts pairwise identity scores in linear time and space complexity.
• Applied to large-scale sequence databases and bacterial genomes.

Main Results:

• Identity achieves linear time and space complexity for predicting identity scores.
• Demonstrates superior speed (2-80x faster) compared to BLAST, Mash, MUMmer4, and USEARCH.
• Exhibits the best performance for low-identity matches and produces the most accurate phylogenetic trees.
• Successfully analyzes millions-of-nucleotides-long bacterial genomes, a feat impossible for global-alignment-based tools.

Conclusions:

• Identity offers a significant advancement in scalable pairwise sequence identity scoring.
• It provides a robust and efficient solution for large-scale genomic data analysis and phylogenetic reconstruction.
• The tool is particularly effective for identifying distant evolutionary relationships and analyzing massive genomic datasets.