<p>Complex prediction within a protein interaction network is essential in cell biology, facilitating the knowledge and enhancement of numerous vital biological functions. The growing accessibility of extensive protein interaction data requires an in-depth analysis to understand cellular architecture and its functions at the network level. Bioinformatics and data mining researchers have examined the functional and structural properties of protein-protein interaction networks using network clustering. In multiple studies over the last two decades, several algorithms have been introduced, ranging from density-based clustering to highly advanced computational methods, each with varying efficacy in identifying overlapping, small, or sparse complexes. The primary objective is to provide new insights into protein complex identification by summarizing existing algorithms and highlighting key approaches, with a focus on their computational efficiency and biological relevance. Unresolved issues related to the field, such as detecting low-density complexes and separating overlapping components, are given special attention. Our review further investigates benchmarking approaches and publicly available PPI tools that promote fairness and reproducibility in evaluation. The key contribution of this review is to offer a structured analysis of clustering-based methods, outlining their strengths, limitations, and practical applications. By consolidating current knowledge, we aim to guide researchers toward efficient, scalable, and biologically meaningful approaches for advancing protein complex prediction.</p>