Hierarchical clustering of gene expression profiles with graphics hardware acceleration

Hierarchical clustering is becoming a de facto standard for analyzing gene expression data. But high computational complexity limits its application in high throughput processing of massive microarray data. An implementation based on commodity graphics hardware is proposed to accelerate this process by employing the parallelism and programmability in graphics pipeline. Significant acceleration is achieved by careful design and implementations, especially in the distance calculation part. The performance comparison between CPU and GPU implementation gives inspiring results.