Performance of genetic programming optimised Bowtie2 on genome comparison and analytic testing (GCAT) benchmarks

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Next-generation sequencing (NGS) technologies generate short, noisy DNA reads.
• Assembling complete DNA sequences relies on matching these reads to reference genomes.
• Current alignment tools are computationally intensive and require specialized tailoring.

Purpose of the Study:

• To develop an automated method for optimizing DNA sequence alignment tools.
• To enhance the speed of the state-of-the-art alignment tool Bowtie2 for NGS data.

Main Methods:

• Utilized genetic programming (GP) to automatically generate optimized versions of alignment algorithms.
• Applied GP to the Bowtie2 tool, creating Bowtie2 (GP).
• Benchmarked Bowtie2 (GP) against the original Bowtie2 using synthetic GCAT datasets.

Main Results:

• Bowtie2 (GP) demonstrated significant speed improvements over the original Bowtie2.
• The enhancements were successfully integrated into the latest Bowtie2 release (v2.2.3).
• Performance was evaluated on both single-ended and paired-end synthetic NGS data.

Conclusions:

• Bowtie2 (GP) achieves up to 45% faster alignment speeds on synthetic NGS benchmarks.
• The trade-off in accuracy is minimal, ranging from 0.2-0.5% and up to 2.5% for longer sequences.