An adaptive grid implementation of DNA sequence alignment

In this paper we have described a dynamic programming algorithm to compute k non-intersecting near-optimal alignments in linear space. In order to reduce its runtime significantly, we are using a hierarchical grid system as the computing platform. Static and dynamic load balancing approaches are investigated in order to achieve efficiently mapping onto this type of architecture, which has characteristics such as: (1) the resources in the grid systems have different computational power; (2) the resources usually are connected by networks with widely varying performance characteristics. At last, a new dynamic load balancing approach named scheduler-worker parallel paradigm is proposed and evaluated.