Cloud computing is an emerging paradigm that provides flexibility, elasticity, high utilization and pay-as-you-go-model while removing the overheard of maintaining a system-specific, fixed-sized cluster of under-utilized machines. On the physical layer, clouds use commodity processors and networking backplanes to build the resources supporting the virtualized cloud. Currently, cloud systems cannot effectively support scientific applications that typically require high-end processors with large bandwidth channels to memory and other processing nodes. Further, software environments in cloud systems are not optimized to target scientific applications. Another limitation could be the mismatch between the scientific applications and the programming model. We plan to evaluate and develop a cloud system to target scientific applications relevant to Qatar and the region. We will then evaluate and analyze the computation and bandwidth requirements of these applications. Further, we will evaluate the utility of the MapReduce programming model for each application. We will design and configure, Qloud, a cloud infrastructure that attempts to match the identified requirements. The performance of executing these scientific applications will be evaluated on Qloud and compared to other cloud and conventional high performance computing (HPC) systems. Further, if the MapReduce model is not applicable, we plan to evaluate other programming models for the chosen applications.