high performance computing on graphics processing units: hgpu.org

Database searching is a main method for protein identification in shotgun proteomics, and many research efforts are dedicated to improving its effectiveness. However, the efficiency of database searching is facing a serious challenge, due to the ever fast growth of protein and peptide databases resulted from genome translations, enzymatic digestions, and post-translational modifications (PTMs). On the other hand, as a general-purpose and high performance parallel hardware, Graphics Processing Units (GPUs) develop continuously and provide another promising platform for parallelizing database searching based protein identification. It becomes very important to study how to speed up database search engines by GPUs for protein identification. In this paper, we mainly utilize GPUs to accelerate the scoring module, which is the most time-consuming component. Specifically, we study two popular scoring method: spectral dot product based method, which is used by X!Tandem, and kernel spectral dot product, which is used by pFind.