We present a computational method for performing structural translation, which has been studied recently in the context of analyzing the steady states and dynamical behavior of massaction systems derived from biochemical reaction networks. Our procedure involves solving a binary linear programming problem where the decision variables correspond to interactions between the reactions of the original network. We call the resulting network a reaction-toreaction graph and formalize how such a construction relates to the original reaction network and the structural translation. We demonstrate the efficacy and efficiency of the algorithm by running it on 508 networks from the European Bioinformatics Institutes' BioModels database. We also summarize how this work can be incorporated into recently proposed algorithms for establishing mono and multistationarity in biochemical reaction systems. Figure 1: A chemical reaction network (left) corresponding to a histidine kinase network where X and Y are two signaling proteins and p is a phosphate group [4]. This CRN has elementary flux modes {r 1 , r 2 , r 4 } and {r 2 , r 3 } which correspond to the directed cycles in the reaction-toreaction graph (center). The structural translation (right) has the same elementary flux modes and stoichiometric vectors as the CRN but the elementary flux modes correspond to cycles.Further connections between the deficiency and the steady states of mass-action systems have been established [11,12,13,14,15,7,6].The study of the deficiency was recently initiated in generalized chemical reaction networks (GCRNs) [31,32]. In a GCRN, each vertex in the reaction graph is associated with two potentially distinct complexes, one for the stoichiometry and one for the kinetic rate of the reaction. Surprisingly, for weakly reversible generalized mass-action systems which have a stoichiometric and kinetic-order deficiency of zero, we still obtain a simple monomial parametrization of the steady state set. A process for relating CRNs and GCRNs, called network translation, was furthermore established in [21]. Network translation consists of restructuring a given CRN in such a way that the resulting network (a GCRN) can be used to guarantee dynamical and steady state properties of the original CRN. The process has been utilized to establish connections between chemical reaction network theory [9], the algebraic study of toric varieties [6,29,8], and biochemical reaction modeling [22,37,5]. Recent work has also established a deficiency-based method for constructing rational parametrizations of steady state sets for a broad class of mass-action systems [23].In this paper, we focus on computational methods for performing the structural component of network translation, which we call structural translation. In general, given a biochemical reaction network of realistic scale, it is challenging to determine a suitable (e.g. weakly reversible, deficiency zero) structural translation. We extend the recent computational work of [22,37] by introducing an elementary flux m...