LAMSA: fast split read alignment with long approximate matches

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• High-throughput sequencing technologies are increasing read lengths, offering potential for advanced genomic studies.
• Longer reads can span structural variant breakpoints, posing challenges for traditional alignment methods.
• Current long-read alignment is inefficient, hindering broad application.

Purpose of the Study:

• To develop a novel split read alignment approach for efficient long-read alignment.
• To address the challenges of aligning long reads with structural variants.
• To improve the speed and accuracy of structural variant detection in genomic data.

Main Methods:

• Proposed LAMSA (long approximate matches-based split aligner), a two-step alignment strategy.
• Initially splits reads into long fragments for co-linear alignment, handling small variations and repeats.
• Employs sparse dynamic programming for split alignment to manage large or non-co-linear variants.

Main Results:

• LAMSA demonstrated substantial speed improvements over state-of-the-art long-read aligners.
• The tool effectively handles various categories of structural variants.
• Performance was validated on simulated and real datasets with diverse read lengths and error rates.

Conclusions:

• LAMSA offers a faster and effective solution for long-read alignment.
• The approach is well-suited for identifying structural variants in genomic studies.
• LAMSA is publicly available, facilitating its adoption in the research community.