cDNA microarrays are a powerful tool for studying gene expression levels. A challenging and complex task of microarray image analysis is the creation of a grid that matches the spots in the image. Proposed methods and tools usually require human intervention, leading to variations of the gene expression results. Furthermore, while automatic methods are available, they present high computational complexity. In this work, the authors present a performance enhancement via GPU computing techniques of an automatic gridding method, previously proposed by their research group. Complex steps of the algorithm were computed in parallel by utilizing the NVIDIA CUDA architecture that allows the use of NVIDIA GPUs for general purpose parallel computations. Experiments showed that the proposed approach achieves higher utilization of the available computational resources, leading to enhanced performance and significantly reduced computational time.
|Title of host publication||Signal Processing and Information Technology(ISSPIT), 2013 IEEE International Symposium on|
|Publication status||Published - 2013|
- cDNA microarray gridding
- GPU computing
- genetic algorithm
Katsigiannis, S., & Maroulis, D. (2013). Parallel Computing Techniques for Performance Enhancement of a cDNA Microarray Gridding Algorithm. In Signal Processing and Information Technology(ISSPIT), 2013 IEEE International Symposium on (pp. 446-451). IEEE. https://doi.org/10.1109/ISSPIT.2013.6781922