<p>Identifying protein complexes from protein-protein interaction networks is one of the crucial tasks in computational biology. Traditional methods, along with their shortcomings in fully understanding protein complex composition, also have inherent limitations and are expensive to implement. In this paper, we introduce a novel method that not only acknowledges but actively tackles these challenges. Our approach, centered around a core-attachment framework, employs a blend of topological metrics, such as square clustering coefficients, in conjunction with traditional clustering coefficients. After establishing the core, we incorporate attachment proteins based on specific conditions employing a depth-first search (DFS) methodology, to form a protein complex. By harnessing multiple metrics, our goal is to elevate the accuracy of protein complex identification beyond what single-metric approaches can achieve. To validate the effectiveness of our approach, we conducted extensive experiments using multiple datasets, including Gavin06, Krogan core, Krogan extend, and DIP datasets, and assessed metrics such as precision, recall, F-measure, and coverage. Our results not only demonstrate the superiority of our method over traditional approaches but also align with findings from related studies. Overall, our study contributes to the ongoing efforts in computational biology by presenting a comprehensive approach to protein complex identification that addresses the shortcomings of previous methods. Through a combination of innovative techniques and insights from recent research, we aim to push the boundaries of accuracy and comprehensiveness in protein complex detection.</p>