SigAlign: an alignment algorithm guided by explicit similarity criteria.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Heuristic aligners in biological sequence alignment offer high throughput via approximation techniques.
• These methods often sacrifice output clarity and introduce complex parameter spaces.
• There is a need for alignment algorithms with transparent criteria and customizable parameters.

Purpose of the Study:

• Introduce SigAlign, a novel alignment algorithm addressing limitations of heuristic methods.
• Provide a transparent and customizable alignment process for bioinformatics tools and pipelines.
• Enhance biological sequence alignment through explicit cutoffs and affine gap penalties.

Main Methods:

• Developed SigAlign, a non-heuristic alignment algorithm.
• Implemented explicit cutoffs: minimum length and maximum penalty per length.
• Incorporated three affine gap penalties.
• Conducted comparative analyses against leading tools: BLASTn, MMseqs2, BWA-MEM, bowtie2, HISAT2, and minimap2.

Main Results:

• SigAlign demonstrates high sensitivity and throughput in comparative analyses.
• Outperforms existing heuristic aligners, especially for high-accuracy reads or genomes with low repetitive content.
• Achieves superior performance in both read mapping and database search tasks.
• SigAlign provides a non-heuristic approach with explicit, understandable output criteria.

Conclusions:

• SigAlign offers a viable alternative to heuristic alignment methods.
• Its transparent and customizable nature makes it suitable for integration into various bioinformatics pipelines.
• The algorithm enhances biological sequence alignment efficiency and clarity.
• SigAlign is released as an open-source library to foster further development and adoption.