|
|
|
motifGPU
GPU-computing
Greatly
Accelerate
Motif
Analysis
|
|
Computational discovery and detection of TF binding patterns has become an indispensible tool in functional genomics
research. Probabilistic approaches utilizing Position-Specific Weight Matrix (PSWM) have demonstrated superior performance; however, they are generally time-consuming. Given that most of the computa-tion time is spent calculating matching scores
position by position, scalable approaches that can take advantage of parallel hardware architectures are highly desirable.
Using NVIDIA Cuda, we have developed a graphics processing unit (GPU)-accelerated motif analysis software library named
motifGPU. Extensive performance comparison studies on two different graphics cards from NVIDIA, (GeForce) GTX 260 and
GeForce GTX 480, showed that commonly-used model-based motif scan procedures can be dramatically accelerated using our
library. De novo motif discovery can also benefit from our GPU-accelerated tools.
|
|
Download program |
Download the latest version of motifGPU library. [August, 2010]
|
|
Documentation |
The manuscript describing motifGPU library.
|
| |
Contact
For bug report, questions or comments, please contact Pooya Zandevakili (zandv@umich.edu).
|
|
