Title: Anisotropic short-range attractions precisely model branched erythrocyte aggregates

Abstract: Homogeneous suspensions of red blood cells (RBCs or erythrocytes) in blood plasma are unstable in the absence of driving forces and form elongated stacks, called rouleau. These erythrocyte aggregates are often branched porous networks -- a feature that existing red blood cell aggregation models and simulations fail to predict exactly. Here we establish that alignment-dependent attractive forces in a system of dimers can precisely generate branched structures similar to RBC aggregates observed under a microscope. Our simulations consistently predict that the growth rate of typical mean rouleau size remains sub-linear -- a hallmark from past studies -- which we also confirm by deriving a reaction kernel taking into account appropriate collision cross-section, approach velocities, and an area-dependent sticking probability. The system exhibits unique features such as the existence of percolated and/or single giant cluster states, multiple coexisting mass-size scalings, and transition to a branched phase upon fine-tuning of model parameters. Upon decreasing the depletion thickness we find that the percolation threshold increases and the morphology of the structures opens up towards an increased degree of branching. Remarkably the system self-organizes to produce a universal power-law size distribution scaling irrespective of the model parameters.