Much attention has been given recently to the development of methods that utilize the large quantity of genetic information available in online databases. Most of the proposed methods look at the entire set of genes and their impact on a disease. Recently a new philosophy emerged which considers using the information available as the genetic pathways, which contain sets of genes and produces a combined effect on a disease. With this newly acquired biological knowledge about pathways the goal is to identify the significant genetic pathways and the corresponding influential genes within a pathway in regard to different diseases. This directly benefits patients by providing targeted treatments based on gene pathway-specific therapies. In this paper we propose a Bayesian kernel machine model which can incorporate existing information on genetic pathways and gene networks in the analysis of DNA microarray data. The overall effect of each pathway is modeled nonparametrically using a reproducing kernel Hilbert space. Multiple pathways are modeled simultaneously using the additive model structure. The pathway and the gene selection within a pathway is done with the help of mixture priors on the pathway indicator variables and the gene indicator variables, respectively. Our proposed kernel-based method provides a very flexible modeling framework to model both linear and nonlinear pathway effects. It can pinpoint the important pathways along with the active genes within each individual important pathways. An efficient Markov chain Monte Carlo algorithm is developed to fit our proposed model. The effectiveness of our method is illustrated though several simulation studies and a real data analysis with right-censored survival outcomes.