One of the biggest questions in cosmology is to understand the physical origin of the late acceleration of the universe. Two classes of models emerge as an attempt to explain this phenomenon: dark energy and modified gravity. These two types of models can reproduce exactly the same expansion history, so it is necessary to go beyond the expansion of the universe and observe the growth of structures to distinguish them. The goal of this thesis is to build observables sensitive to the growth of structures and therefore to use them to test gravity. This can be done in both linear and non-linear regimes. In the linear regime, we propose two tests, which target two different properties of the theory of gravity. In the non-linear regime, we use a new observable, called the line correlation function, to extract information beyond the power spectrum and improve the tests of gravity.