Neural-symbolic computing has now become the subject of interest of both\nacademic and industry research laboratories. Graph Neural Networks (GNN) have\nbeen widely used in relational and symbolic domains, with widespread\napplication of GNNs in combinatorial optimization, constraint satisfaction,\nrelational reasoning and other scientific domains. The need for improved\nexplainability, interpretability and trust of AI systems in general demands\nprincipled methodologies, as suggested by neural-symbolic computing. In this\npaper, we review the state-of-the-art on the use of GNNs as a model of\nneural-symbolic computing. This includes the application of GNNs in several\ndomains as well as its relationship to current developments in neural-symbolic\ncomputing.\n