Bioinformatics Methods
Current understanding
Short-read alignment sits at the foundation of most modern genomic pipelines. The Burrows-Wheeler Aligner (BWA), introduced by Li & Durbin in 2009, replaced the then-dominant MAQ by exploiting the Burrows-Wheeler transform to achieve 6–18× faster alignment against the human genome without sacrificing accuracy. The speedup scales with read length: MAQ internally pads all reads to 128 bp regardless of true length, so its runtime is insensitive to actual read length, while BWA’s BWT-based approach scales more favorably. That architectural difference is what drives the performance gap — not simply tighter code.
This kind of benchmark matters because alignment is rarely the only computationally expensive step in a workflow, but it is almost always the earliest bottleneck on raw data. Choosing an aligner that is an order of magnitude faster with comparable error rates compresses wall-clock time across entire projects, not just a single run.
Supporting evidence
- Fast and accurate short read alignment with Burrows-Wheeler transform., Finding 1 — BWA is 6–18× faster than MAQ on short reads mapped to the human genome, with the range depending on read length.
Contradictions / open disagreements
The published abstract reports a 10–20× speedup while the results section of the same paper reports 6–18×. The discrepancy tracks which read-length range is used in the comparison. More broadly, the benchmarks were conducted on a single 2.5 GHz Xeon E5420 core with simulated data; whether the same gains hold on modern multi-core hardware with real sequencing reads and more complex or repeat-rich genomes has not been settled by this paper alone.
Tealc’s citation-neighborhood suggestions
Papers benchmarking BWA against Bowtie, SOAP2, and later tools (e.g., BWA-MEM) on real datasets across diverse genome sizes would sharpen the picture of where the BWT-based approach excels versus where seed-and-extend or other strategies close the gap.