Cancer Biomarker Discovery Using High Performance Algorithm

Biomarker has an important role for early diagnosis and also it can predict progression disease and provide monitoring treatment. Biomarker is a structure in the body that has biological changes due to a specific disease. Analyze nucleic acids, protein and metabolism may provide cancer biomarkers. Biomarkers discovery and study of their status is possible using bioinformatics tools. However, Due to high-dimensiona data and several types of them along with lack of specific method for biomarker discovery, choosing the biomarkers that have diagnostic and predictive information is not easy. In this paper, we introduce bioinformatic tools for biomarker discovery such as datasets and computational tools then we focus on feature selection algorithms as a house for biomarker discovery. Using feature selection algorithms in addition to the given a subset of important and relevant features, reduce data dimensional and increase accuracy. In this way we introduce the best feature selection methods and classification algorithms used for this purpose and discuss computational challenges such as performance and stability. In this research we select breast cancer and provide data from The Cancer Genome Atlas (TCGA) data portal that include 1800 patients and more than 2500 attributes. At the end of this paper we introduce parallel computing as a solution for high-dimensional data analysis.

کلیدواژه ها:

Biomarker ، Feature Selection algorithms ، Parallel Computing ، Breast Cancer ، TCGA