Computational Sequence Design Techniques for DNA Microarray Technologies

Introduction

The design of DNA oligos is a key step in the manufacturing process of modern microarrays – biotechnology tools that allow the parallel qualification and quantification of large numbers of genes. Areas that have benefited from the use of microarrays include gene discovery (Andrews et al., 2000; Yano, Imai, Shimizu, & Hanashita, 2006), disease diagnosis (Yoo, Choi, Lee, & Yoo, 2009), species identification (Pasquer, Pelludat, Duffy, & Frey, 2010; Teletchea1, Bernillon, Duffraisse, Laudet, & Hänni, 2008) and toxico-genomics (Jang, Nde, Toghrol, & Bentley, 2008; Neumanna and Galvez, 2002).

Microarrays consist of plastic or glass slides, to which a large number of short DNA sequences (probes) are affixed at known positions in a matrix pattern. A probe is a relatively short DNA sequence (20-70 bases) representing the complement of a contiguous sequence of bases from a target that acts as its fingerprint. The purpose of each probe is to uniquely identify and bind a target via a process called hybridization. Nevertheless, in practice probes could bind to more than one target via a process called cross-hybridization.

While microarrays could be used for a variety of applications like transcription factor binding site identification (Hanlon & Lieb, 2004), eukaryotic DNA replication (MacAlpine & Bell, 2005), and array comparative genomics hybridization (Pinkel & Albertson, 2005), their main use remains gene transcript expression profiling (Schena, Shalon, Davis, & Brown, 1995; Ross et al., 2000; Aarhus, Helland, Lund-Johansen, Wester & Knappskog, 2010). However, at present, the fundamental understandings of the bio-chemo-physical mechanisms that power this technology are poorly understood (Pozhitkov, Tautz, & Noble, 2007), thus leading to hybridization signal levels that are still not accurately correlated with exact amounts of target transcripts. While most of the microarray research work carried today focuses on the development of reliable and fault-tolerant statistical techniques that could pre-process large data sets (Holloway, van Laar, Tothill, & Bowtell, 2002; Irizarry et al., 2003; Quackenbush, 2002; Yang et al., 2002; Zhao, Li, & Simon, 2005) and identify significant factors relevant to each particular study (Chu, Ghahramani, Falciani, & Wild, 2005; Harris & Ghaffari, 2008; Leung & Hung, 2010; Peng, Li, & Liu, 2007; Zou, Yang, & Zhu, 2006), more work needs to be done on improving the infrastructural aspects of microarray technology, thus reducing the amount of noise earlier rather than later in an experiment based on microarrays data.

Thus, one of the greatest challenges in DNA microarray design resides in how to select large sets of unique probes that distinguish among specific sequences from complex samples consisting of thousands of closely similar targets. The daunting task of designing such large sets of probes is hampered by the computational costs associated with probe efficacy evaluations. Various design strategies are presented that employ the utilization of intricate probe evaluation criteria. Some of these strategies were inspired from design techniques employed for solving similar problems that arise in coding theory (Bogdanova, Brouwer, Kapralov, & Östergård, 2001; Gaborit & King, 2005; Gamal, Hemachandra, Shperling, & Wei, 1987), bio-molecular computing (Feldkamp, Banzhaf, & Rauhe, 2000; Frutos et al., 1997), molecular tagging (Braich et al., 2003; Brenner & Lerner, 1992) and nano-structure design (Reif, Labean, & Seeman, 2001; Yurke, Turberfield, Mills, Simmel, & Neumann, 2000).