Metabolic networks used for quantitative analysis often contain links that are irrelevant for graph-based structural analysis. For example, inclusion of side compounds or modelling artifacts such as 'biomass' nodes. While Carbon Skeleton Graph offer a relevant alternative topology for graph-based analysis, it requires compounds' structure information, usually not provided in model, and difficult to retrieve for model with sparse cross-reference annotations. In contrary to the SBML2Graph app that performs a raw conversion of the SBML content, the present app propose a fine-tuned creation of compound graph from predefined list of side compounds and degree² weighting to get relevant structure without structural data.This app also enable Markov-chain based analysis of metabolic networks by computing reaction-normalized transition probabilities on the network.