Essential protein identification is an important factor to inspect the mechanisms of disease progression and to identify drug targets. With the advancement of high throughput genome sequencing projects, a bulk of protein data is available where the analysis of interaction pattern, functional annotation and characterization are necessary for detecting proteins' essentiality in network level. A set of centrality measure has been used to identify the highly connected proteins or hubs. From recent studies, it is observed that the majority of hubs are considered to be essential proteins. In this article, a method EPIN_Pred is proposed where a combination of several centrality measures is used to find the hub and non-hub proteins. Using the cohesiveness property, overlapping topological clusters are found. Using gene ontology (GO) terms, these topological clusters are again combined, if required. The performance of EPIN_Pred is also found to be superior when compared to other state-of-the-art methods.