The era of genome sequencing and the development of statistical genetics have linked numerous genetic variants to various human diseases and traits. Understanding the causal effects of these variants can uncover the specific biological pathways underpinning each phenotype, thereby informing the development of novel therapies. This dissertation addresses the challenges of functionally interpreting variants, highlighting from distinct perspectives the importance of considering when and where in exploring the mechanistic contribution of a variant to diseases or traits. In the first study, we focus on a group of genetically-linked variants associated with an increased risk of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). We establish that these variants increase disease risk by promoting the inclusion of a detrimental cryptic exon in a critical neuronal gene, UNC13A, only in the pathological condition. In the second study, we turn to variants associated with congenital heart defects (CHD) and quantitative cardiac traits. Through the development of a gene regulatory atlas of the human fetal heart, we find that certain common noncoding variants affect enhancers highly specific to particular sub-anatomical heart structures. Furthermore, the atlas reveals that rare coding variants and common noncoding variants associated with related phenotypes converge on similar biological pathways, suggesting that disease severity caused by rare coding variants can be modulated by common noncoding variants in the background. Together, this dissertation argues that only by incorporating contextual information, such as disease state, cell type, and genetic background, can we effectively identify the biological mechanisms underlying the association between the variants and the phenotypes.