Converting Biological Neural Networks to DAGs: Evaluation of Customized Algorithms on C. Elegans

With the increase of available data on biological connectomes, the ability to comprehend and utilize these networks becomes progressively important. In this paper, we customize depth-first search (DFS) and minimum feedback arc set (MFAS) algorithms to remove cycles within the C. elegans adult hermaphrodite connectome. Furthermore, we evaluate the performance of these algorithms using multiple metrics surrounding the architecture, including the edges, the sensory neuron to neuromuscular junction pathways, and other graphical properties of the connectome. The results indicate that the MFAS algorithm retains the structural authenticity of the original graph significantly better than the DFS algorithm.