GROOLS (Genomic Rule Oriented Object Logic System) is an expert system to help biologists in the evaluation of genome functional annotation through biological processes like metabolic pathways. Such reasoning is conducted using a Business Rule Management System (BRMS) working on a generic representation of biological knowledge that captures complex data and relationships. We use the Object-Oriented First Order Logic (OOFOL) extended to a four-valued logic of Belnap. Prior biological knowledge is organized in a directed acyclic graph and evaluated by applying reactive graph reasoning using observation facts. Two types of observations are considered: predictions from bioinformatics methods and assertions that correspond to experimental evidences in the studied organism. Once all facts are spread over the graph, a conclusion is made about the state of prior knowledge (e.g. confirmed presence, missing, unexpected absence). GROOLS implementation is based on the jBoss DROOLS framework.