Clustering of proteins is an essential as it helps to infer biological function of a new sequence. In this paper, the protein sequences of Mycobacterium Tuberculosis have been clustered based on its space group using Refined Markov Clustering algorithm. The proposed technique reduces the overlapping clusters and performs better than other algorithms. This approach minimizes the proceeding time for the protein sequence effectively. The proposed work was evaluated by com¬¬parative analysis with k-medoids, spectral normalized cut and fast connected component algorithm. According to the clustering validation and comparison results, the proposed algorithm performs better than other algorithms.