In the field of bioinformatics, small to large data sets of genes, proteins, and genomes are analyzed for biological significance. A technology that has been in the forefront of generating large amounts of gene data is the microarray or hybridization technique. It has been instrumental in the success of the human genome project and paved the way for a new era of genetic screening, testing, and diagnostics (Scheena, 2003). The microarray data set can be made of thousands of rows and columns. It often contains missing values, exhibits high-dimensional attributes, and is generally too large for manual management or examination (Tseng & Kao, 2005; Turner, Bailey, Krzanowski, & Hemingway, 2005). Database technology is necessary for the extraction, sorting, and analyzing of microarray data sets.
Numerous data mining studies based on partitioning groups in multidimensional microarray data sets reveal the significance of clustering in biological information extraction (Au, Chan, Wong, & Wang, 2005; Bolshakova & Azuaje, 2006; Lacroix, 2002; Piatetsky-Shapiro & Tamayo, 2003). A broad overview of biostatistical clustering approaches in microarray analysis is given by Scheena (2003). There are large clustering algorithms in the literature, but they are relatively equivalent in performance.
Key Terms in this Chapter
Centroid: The centroid of a cluster is the average point in the multidimensional space defined by the dimensions.
K-Means Clustering: Objects are grouped into a fixed number (k) of partitions so that the partitions are dissimilar to each other.
Hierarchical Clustering: This involves the recursive clustering of data points, which may be agglomerative or divisive. An agglomerative clustering method starts with each case in a separate cluster and then combines the clusters sequentially, reducing the number of clusters at each step until only one cluster is left.?The divisive hierarchical clustering starts with all objects in one cluster and then subdivides them into smaller pieces.
Dendrogram: A two-dimensional diagram illustrating the fusions or divisions made at each successive stage of hierarchical clustering.
Microarray Analysis: It involves five-step exploratory approach to analyzing microarray data. The first step involves posing the biological question the experiment should address. Sample preparation is the next step, which includes experiments to be performed such as DNA/ RNA isolation, probe amplification, and microarray manufacture. The next step is the biochemical reactions between the target and probe molecules, for example, hybridization. The fourth step detects captured images from the microarray using scanners or imaging instruments. Finally, the captured images are analyzed and modeled.
Microarray Data: This is a data set in matrix form containing gene expression values of samples. The row entry represents a gene while the column contains the chip.
Filtering: The process of removing genes that are not expressed or do not vary across sample types.
Self-Organizing Maps (SOMs): A neural-network method that reduces the dimensions of data while preserving the topological properties of the input data. SOM is suitable for visualizing high-dimensional data such as microarray data.
Bioinformatics: A broad term used to describe the collection, organization, and analysis of biological data using computer technology and information science. Some bioinformatics activities include the modeling of protein structure and creation of extensive electronic databases on genes, genomes, and protein sequences.