Biological Data Prediction Using Two Mode Grouping Bayesian Principal
M. Sangeetha1, P. Bhuvaneswari2, A. Sujitha3, P. Nandhini4, C. Gurulakshmi5
1Ms. M. Sangeetha, IT Department, Sri Krishna College of Technology, Coimbatore, India.
2Ms. P. Bhuvaneswari, IT Department, Sri Krishna College of Technology, Coimbatore, India.
3Ms. A. Sujitha, IT Department, Sri Krishna College of Technology, Coimbatore, India.
4Ms. P. Nandhini, IT Department, Sri Krishna College of Technology, Coimbatore, India.
5Ms. C. Gurulakshmi, IT Department, Sri Krishna College of Technology, Coimbatore, India.
Manuscript received on January 20, 2015. | Revised Manuscript received on February 05, 2015. | Manuscript published on February 28, 2015. | PP: 203-207 | Volume-4 Issue-3, February 2015. | Retrieval Number: C3795024315/2013©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The development of DNA chip technology makes it possible that high-throughput gene expression profiles could be observed simultaneously in particular living organism. The obtained data are usually shown in the form of matrix with genes in rows and experimental conditions in columns. However, these matrices often contain missing values caused by various factors, such as hybridization failures, insufficient resolution, or deposition of dust or scratches on the slide. The subsequent analyses of gene expression data (e.g. clustering, inferring regulatory model, or finding functional gene) always require the complete matrices. Repeating the experiments to obtain a complete gene expression matrix is usually costly and unpractical. Omitting the gene expression profile vector with missing values may lose useful information. Substituting the missing values with zeros or row averages lead the change of variance among variables. So an efficient imputation method for the missing value is needed.
Keywords: DNA Chip, Hybridization, Clustering, Genes.