Imaging genetics has advanced to uncover genetic variants associated with imaging phenotypes characterized by imaging data, which can be considered functional. Functional data refers to data that are inherently functions, where each observation represents a curve, or some other form of continuous variation over a domain.In this study, we propose a novel approach, the functional sequence kernel association test (fSKAT), aimed at identifying single-nucleotide polymorphisms (SNPs) associated with functional phenotypes. To address the infinite space of functional data, we employ functional principal component analysis to extract key features, specifically functional principal component scores (fPC scores), from the functional phenotype. FSKAT extends multi-SKAT, which has been to incorporate non-linear SNP effects on multivariate phenotypes, leveraging fPC scores as a multivariate response. Our focus lies on resting-state functional connectivity, which represents the synchronized activity between different brain regions in the absence of any task. We construct the functional phenotype as a functional connectivity curve (FC curve), depicting the trajectory of a seed brain region's functional connectivity alongside other regions. The FC curve, defined by the Euclidean distance along its domain and functional connectivity as its values, enables us to consider spatial correlations among adjacent brain regions within functional connectivity.Through a comprehensive simulation study, our method demonstrates superior statistical power compared to competing approaches that do not account for the functional nature of imaging data. In our real data analysis, we can pinpoint significant SNPs associated with disrupted functional connectivity in mild cognitive impairment patients at high risk of Alzheimer's disease.