Many diseases are influenced by the marginal effects of genetic covariates (G) and environmental covariates (E), as well as their interactions. These interactions are most commonly addressed by adding the GxE terms to the model. However, including the GxE terms may complicate the model, especially when the dimension of G is large. Furthermore, GxE only captures one specific type of interactions, whereas the true interactions can be more general. In this project, we propose a two-stage model as a solution to the aforementioned problems. In the first stage, we calculate the conditional percentile for each individual, adjusting for all the E factors with a global quantile regression model. In the second stage, we select G factors that are associated with the conditional percentile. By modeling the impact of genes and the environment separately in two stages, our proposed method is can identify associated gene markers that potentially have complex interactions with the environment.