In this paper, we study the problem of designing routes for source coding with explicit side information (i.e., with side information at both the encoder and the decoder) in sensor networks. Two difficulties in constructing such data-centric routes are the lack of reasonably practical data aggregation models and the high computational complexity resulting from the coupling of routing and in-network data fusion. Our data aggregation model is built upon the observation that in many physical situations the side information providing the most coding gain comes from a small number of nearby sensors. Based on this model, we formulate an optimization problem to minimize the communication cost, and show that finding the exact solution of this problem is NP-hard. Subsequently, two suboptimal algorithms are proposed. One is inspired by the balanced trees that have small total weights and reasonable distance from each sensor to the fusion center . The other separately routes the explicit side information to achieve cost minimization. Bounds on the worst-case performance ratios of two methods to the optimal solution are derived for a special class of rate models, and simulations are conducted to shed light on their average behaviors.