Arioc: GPU-accelerated alignment of short bisulfite-treated reads

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Aligning bisulfite-treated DNA sequences (BS-seq) to large genomes presents a significant computational challenge.
• This bottleneck impacts the efficiency of BS-seq data analysis.

Purpose of the Study:

• To improve the speed and sensitivity of BS-seq alignment through algorithmic and software engineering enhancements.
• To evaluate a strategy of modifying read-alignment algorithms by integrating BS-seq specific logic.

Main Methods:

• The Arioc short-read aligner, utilizing GPU hardware, was modified to incorporate BS-seq computational logic into both GPU and CPU code.
• Performance was evaluated by comparing Arioc's alignments with those from established CPU-based BS-seq aligners using simulated and human sequencing reads.

Main Results:

• Arioc demonstrated equal or superior accuracy compared to other evaluated BS-seq aligners on simulated reads.
• For human sequencing reads, Arioc achieved a throughput at least 10 times greater than existing aligners across various sensitivity settings.

Conclusions:

• The integration of BS-seq logic into the Arioc aligner effectively mitigates computational bottlenecks.
• Arioc offers a significant performance improvement for BS-seq data analysis, providing a faster and highly accurate alignment solution.