Current Bioinformatics - Volume 2, Issue 3, 2007

Gene expression data are essential in understanding gene functions, biological networks, and cellular conditions. Over the last decade, with the use of DNA microarray technology, larger-scale gene expression data has become available and various data mining techniques have been used in an attempt to extract biologically relevant knowledge. Rule mining is a widely used approach in data mining and lately has attracted considerable interest in Bioinformatics with a number of methods being proposed over the last few years. Rule discovery can reveal biologically relevant associations between different genes or between experimental conditions and gene expression. The widely used data clustering techniques that can successfully identify co-expressed genes can be seen as a special case of rule mining. Furthermore, association rule mining can reveal temporal patterns of gene co-expression inferring participation in gene networks. This paper is a review of recent developments and current state of association rule mining methodologies in gene expression data. We discuss the advantages and limitations of the existing techniques and moreover, we propose a general framework under which association rule mining in gene expression can infer very often meaningful biological results.

Protein structures are classically described as composed of two regular states, the α-helices and the β-strands and one non-regular and variable state, the coil. Nonetheless, this simple definition of secondary structures hides numerous limitations. In fact, the rules for secondary structure assignment are complex. Thus, numerous assignment methods based on different criteria have emerged leading to heterogeneous and diverging results. In the same way, 3 states may over-simplify the description of protein structure; 50% of all residues, i.e., the coil, are not genuinely described even when it encompass precise local protein structures. Description of local protein structures have hence focused on the elaboration of complete sets of small prototypes or “structural alphabets”, able to analyze local protein structures and to approximate every part of the protein backbone. They have also been used to predict the protein backbone conformation and in ab initio/ de novo methods. In this paper, we review different approaches towards the description of local structures, mainly through their description in terms of secondary structures and in terms of structural alphabets. We provide some insights into their potential applications.

Various gene expression DNA microarray platforms have been developed and often similar experiments are conducted using different types of arrays. Several investigators have explored the cross-platform comparison issues and reached mixed conclusions, ranging from discrepant to reasonably concordant results across platforms. These comparative studies, however, did not apply rigorous sequence-matching criteria. Since the sample size in each experiment is small, combining the information may help increase statistical power. Here we address this issue using the probe sequences to determine matching probes. Smooth muscle cells from the thoracic aorta of C57B46J mice treated with benzo(a)pyrene or dimethyl sulfoxide were examined using GeneChip and CodeLink arrays. Corresponding probes on the two platforms were identified by stand-alone BLAST. In general, CodeLink selected more genes as differentially expressed. For GeneChip arrays, among the various algorithms considered, more probe sets were found significantly changed using the MAS 5.0 expression estimates. The Li-Wong PM-only expression estimates led to fewer discordant results between the two platforms. We conducted quantitative RT-PCR assays on several genes to further assess the performance of each platform. We found that the use of different expression estimation algorithms led to different results within the same platform. The qRT-PCR experiments did not consistently confirm the results of either platform.

Molecular characterization of various microbial genomes has revealed that many pathogenic and mutualistic intracellular bacterial species have smaller genomes than their free-living relatives. The flux, streamlining and elimination of genes in these genomes constitute a selective and ongoing process. Obligatory intracellular parasites display marked similarities in patterns of protein length and frequency distribution, with substantial sharing of a ‘backbone genome’. The results highlight that gene loss is function-dependent, but is independent of protein length. It is suggested that obligate genomes have greater proportion of overlapping genes, which may be a result of evolutionary pressure to minimize genome size. The differential loss of genes in these organisms is also reflected in their metabolic plasticity. Here, we present a review on gene loss and gene conservation in obligatory intracellular parasites. These data are essential for understanding of genome evolution and microbial pathogenesis of these organisms.

Ets family transcription factors, characterized by an evolutionary conserved Ets domain, play important roles in cell development, cell differentiation, apoptosis and tissue remodeling. All members of this family have an activation or a repression domain for DNA binding. The Ets domain has been shown to bind 5'-GGAA/T-3' core motif of DNA by its wHTH (winged helix-turn-helix) structure, as determined by NMR or X-Ray crystallography analysis. A number of Ets transcription factors have been shown to be involved or directly implicated in the pathogenesis of a wide spectrum of human cancers. They are thus potential molecular targets for selective cancer therapy. Computer simulations using molecular dynamics allows monitoring the dynamics of individual atoms, thereby giving a unique insight into the structural flexibility of binary complexes between Ets-domain and a specific DNA sequence that cannot be easily extracted from laboratory experiments. Moreover, an understanding of structural motifs that influence the specificity of Ets domains is essential for antitumor drug design. Here we describe molecular and structural studies that provide a detailed description of the direct and indirect readout mechanisms of DNA recognition by Ets domains, a conserved mechanism observed in other DNA-protein complexes.