Precise control of neuronal pathways is of critical importance for the efficacy of neuro-modulation prostheses designed to restore vision in patients with retinal degeneration. We present a biologically-informed computational framework that enables the generation of pathway-specific spiking patterns for RGCs. The model is used to generate control signals for implantable optical retinal neurostimulator device to modulate genetically-modified RGCs, thus enable cell-type-specific neural excitation. The model is used to compare electrical and optical neuro-stimulations with respect to the quality of visual perception they induce in the brain for the same visual stimulus which shows significantly better results for optical stimulation.