In this dissertation we propose methodology for testing SNP-sets for genetic associations, both for sequencing and genome-wide association studies. Due to the large scale of this kind of data, there is an emphasis on producing methodology that is not only accurate and powerful, but also computationally efficient.