Distributed fuzzy clustering techniques for cancer diagnosis with protein sequence data

Abstract

The cell of a human body has genes in De oxyribo Nucleic Acid (DNA) which controls the functions of the cell by making proteins. The mutation of one or more genes in a cell causes cancer. This mutation causes the abnormal protein creation which has different information in nucleotide sequence, protein structure than a normal protein. Forming a chain of amino acids is called protein. Proteins are utilized in the early stage of cancer detection because the protein sequences and protein structure motif establish amino acid sequence patterns in all parts of a protein. Biomedical data analysis reveals that the protein data is non-numerical in nature. Data mining plays a major role in extracting the information from big volumes of non-numerical protein sequences. The data can be grouped based on similarities which enables the speedy convergence for clustering different sizes of data. Usually, analyzing the huge volume of data requires more computation time for identifying cancerous protein sequences during the diagnosis of cancer disease. The current research utilizes distributed clustering to deal with the computational time complexity issue by distributing the huge data into different local clusters. Additionally, it utilizes the soft computing approach to group similar protein sequences into specific clusters by finding the global clustering from the results of local clusters. The uncertainty of protein sequences is addressed with the utilization of a fuzzy logic system. This, in turn, leads to achieve the main objective of the research which is identifying the cancerous protein sequences at an early stage from the distributed large data and to improve accuracy, F-measure, precision and recall in clustering the protein sequences with the reduction of the False Positive Rate (FPR) as well as newline