Detecting protein complexes is a pivotal task in computational biology, essential for unraveling cellular mechanisms and advancing drug discovery. Despite the demonstrated potential of evolutionary algorithms (EAs) in identifying protein complexes from protein-protein interaction (PPI) networks, their integration with gene ontology (GO) annotations remains underexplored. This study introduces two key contributions: a novel single-objective optimization model for protein complex detection and a new GO-based mutation operator. This operator leverages GO annotations to enhance the synergy between optimization techniques and biological insights, improving algorithmic performance. To our knowledge, this is the first approach to embed PPI biological characteristics into both problem formulation and mutation operator design. The proposed single-objective EA was evaluated on three widely used PPI networks and benchmark datasets. Experimental results demonstrate that our algorithm surpasses state-of-the-art methods in accurately detecting protein complexes, with the proposed heuristic GO-based mutation operator significantly enhancing the quality of results compared to other EA-based approaches.