Computational Intelligence Algorithms and DNA Microarrays
In this chapter, we present Computational Intelligence algorithms, such as Neural Network algorithms, Evolutionary Algorithms, and clustering algorithms and their application to DNA microarray experimental data analysis. Additionally, dimension reduction
- PDF / 352,355 Bytes
- 31 Pages / 439.37 x 666.142 pts Page_size
- 61 Downloads / 183 Views
2
Institute for Mathematical Sciences, Imperial College London, South Kensington, London SW7 2PG, United Kingdom [email protected] Computational Intelligence Laboratory, Department of Mathematics, University of Patras Artificial Intelligence Research Center (UPAIRC), University of Patras, GR–26110 Patras, Greece {vpp,vrahatis}@math.upatras.gr
Summary. In this chapter, we present Computational Intelligence algorithms, such as Neural Network algorithms, Evolutionary Algorithms, and clustering algorithms and their application to DNA microarray experimental data analysis. Additionally, dimension reduction techniques are evaluated. Our aim is to study and compare various Computational Intelligence approaches and demonstrate their applicability as well as their weaknesses and shortcomings to efficient DNA microarray data analysis.
1.1 Introduction The development of microarray technologies gives scientists the ability to examine, discover and monitor the mRNA transcript levels of thousands of genes in a single experiment. The development of technologies capable to simultaneously study the expression of every gene in an organism has provided a wealth of biological insight. Nevertheless, the tremendous amount of data that can be obtained from microarray studies presents a challenge for data analysis. This challenge is twofold. Primarily, discovering patterns hidden in the gene expression microarray data across a number of samples that are correlated with a specific condition is a tremendous opportunity and challenge for functional genomics and proteomics [1–3]. Unfortunately, employing any kind of pattern recognition algorithm to such data is hindered by the curse of dimensionality (limited number of samples and very high feature dimensionality). This is the second challenge. Usually to address this, one has to preprocess the expression matrix using a dimension reduction technique [4] and/or to find a subset of the genes that correctly characterizes the samples. Note that this is not similar to “bi-clustering”, which refers to the identification of genes that exhibit similar behavior across a subset of samples [5, 6]. In this chapter we examine the application of various Computational Intelligence D.K. Tasoulis et al.: Computational Intelligence Algorithms and DNA Microarrays, Studies in Computational Intelligence (SCI) 94, 1–31 (2008) c Springer-Verlag Berlin Heidelberg 2008 www.springerlink.com
2
D.K. Tasoulis et al.
methodologies to face problems arising from the twofold nature of the microarray data. We also examine various manners to combine and interact algorithms towards a completely automated system. To this end the rest of this chapter is structured as follows. Sections 1.2 and 1.3 are devoted to a brief presentation of Neural Networks as classification tools, Evolutionary Algorithms that can be used for dimension reduction, and their synergy. In Section 1.4 various feature selection and dimension reduction techniques are presented, starting from the Principal Component Analysis, continuing with several
Data Loading...
