I-BAR proteins are located at the plasma membrane to sense and generate local membrane curvature. Prior to inducing membrane shape changes larger than a single protein, I-BAR domains are believed to aggregate and assemble. To discern how I-BAR domains organize and collectively stabilize membrane deformation, I developed two distinct coarse-grained membrane-protein models: one is a bottom-up model parameterized from all-atom simulations to capture low bound protein density behavior of the I-BAR domain of IRSp53 and the other is a tunable, lower resolution model used to assess the effects of various characteristics of the I-BAR family (e.g., intrinsic protein curvature). The separate approaches highlight the strengths of various coarse-graining approaches while providing key insights into I-BAR domain assembly. Together, I elucidate the role of attractive membrane-mediated forces in I-BAR domain assembly and the interplay between protein and membrane curvatures. After using both bottom-up and top-down methodologies to understand I-BAR domain self-assembly, I assess how coarse-grained representation affects the estimation of the membrane bending modulus and the temperature-dependence in the popular top-down model, MARTINI. Issues of transferability and representability have been identified in the bottom-up coarse-grained models, but a computational demonstration of these effects in top-down coarse-grained models does not exist. I use MARTINI as a test case to understand how the coarse-grained representation affects cholesterol and amphipathic helix association in bilayers. The correctness of MARTINI in capturing these various effects is of growing interest as coarse-grained simulations are applied outside of parametrization data.