Alternative splicing analysis represents a typical computing pattern in bioinformatics. In such a problem?s domain, there are massive and irregular batch jobs. It is critical to improve such a computation?s efficiency and reduce the cost. Based on computational Grid technologies, a solution has been devised. The solution consists of a Grid service model for representing heterogeneous resources, a WS Service-Group for dynamically discovering as many appropriate resources as possible, and a WS Service- Group for performing massive, irregular batch jobs on distributed resources in a parallel and optimized way. The solution has been implemented in Harmonia, a computational Grid software platform developed by Peking University. Experiments has approved the solutions feasibility and practicability, also has presented some hints for improving its efficiency.