Efficient Protein Alignment Algorithms Using GPUs
Author Information
Author(s): Blazewicz Jacek, Frohmberg Wojciech, Kierzynka Michal, Pesch Erwin, Wojciechowski Pawel
Primary Institution: Poznań University of Technology
Hypothesis
Can GPU-based algorithms improve the efficiency of protein sequence alignment?
Conclusion
The study demonstrates that GPU-optimized algorithms can efficiently compute pairwise alignments, significantly outperforming traditional CPU methods.
Supporting Evidence
- The implementation achieved performance up to 6.3 GCUPS on a single GPU.
- The algorithm can compute pairwise alignments, not just scores.
- Multi-GPU support allows for nearly linear speed increases.
Takeaway
This study shows that using powerful graphics cards can make comparing protein sequences much faster and easier.
Methodology
The study implemented GPU-based dynamic programming algorithms for pairwise sequence alignment, optimizing for backtracking procedures.
Limitations
The algorithm's performance may decrease with longer sequences due to memory constraints.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website