We propose a general framework to perform gene/region based analysis of sequencing data by regressing a functional response on one or multiple scalar predictors. Next generation sequencing technologies make it possible to uncover genetic information from millions of variants. Since the observed sequenced variants are very close in their genetic positions, we can consider them to be realizations of random continuous functions. Therefore, instead of analyzing multiple individual genetic variants per subject, we can estimate the underlying continuous function and treat it as a functional response in a regression model. Smoothing splines are used to fit these functional responses by maximizing the penalized likelihood. Covariates can also be incorporated in the analysis to control for confounding, including qualitative and quantitative predictors. By utilizing a connection between penalized spline regression and linear mixed models, we are able to fit our model using standard linear mixed models statistical packages. To illustrate our approach, we conduct simulation studies and apply our proposed methodology to sequencing data from the Dallas Heart Study.