Fusion of functional magnetic resonance imaging (fMRI) and genetic information is becoming increasingly important in biomarker discovery. These studies can contain vastly different types of information occupying different measurement spaces and in order to draw significant inferences and make meaningful predictions about genetic influence on brain activity; methodologies need to be developed that can accommodate the acute differences in data structures. One powerful, and occasionally overlooked, method of data fusion is canonical correlation analysis (CCA). Since the data modalities in question potentially contain millions of variables in each measurement, conventional CCA is not suitable for this task. This paper explores applying a sparse CCA algorithm to fMRI and genetic data fusion.