A mixture model approach to sample size estimation in two-sample comparative microarray experiments.

BACKGROUND: Choosing the appropriate sample size is an important step in the design of a microarray experiment, and recently methods have been proposed that estimate sample sizes for control of the False Discovery Rate (FDR). Many of these methods require knowledge of the distribution of effect sizes among the differentially expressed genes. If this distribution can be determined then accurate sample size requirements can be calculated. RESULTS: We present a mixture model approach to estimating the distribution of effect sizes in data from two-sample comparative studies. Specifically, we present a novel, closed form, algorithm for estimating the noncentrality parameters in the test statistic distributions of differentially expressed genes. We then show how our model can be used to estimate sample sizes that control the FDR together with other statistical measures like average power or the false nondiscovery rate. Method performance is evaluated through a comparison with existing methods for sample size estimation, and is found to be very good. CONCLUSION: A novel method for estimating the appropriate sample size for a two-sample comparative microarray study is presented. The method is shown to perform very well when compared to existing methods.

Transcriptional responses of Arabidopsis thaliana ecotypes with different glucosinolate profiles after attack by polyphagous Myzus persicae and oligophagous Brevicoryne brassicae.

Plants are equipped with a range of defence mechanisms against herbivorous insects. In cruciferous species, jasmonic acid, salicylic acid, and ethylene along with glucosinolates and their hydrolysis products play important roles in plant protection and plant-insect communication. In turn, a number of herbivores have adapted to plants that contain glucosinolates. As a result of adaptation to their host plants, specialized insects may elicit different plant-inducible responses than generalists. Oligonucleotide microarrays and qRT-PCR analysis were used to characterize transcriptional profiles of Arabidopsis thaliana plants in response to infestation with a generalist aphid, Myzus persicae, or a cruciferous plant specialist, Brevicoryne brassicae. To find possible differences and similarities in molecular responses between plants differing in predominant glucosinolate hydrolysis products, three ecotypes of A. thaliana were chosen: Wassilewskija (Ws), Cape Verde Islands (Cvi), and Landsberg erecta (Ler), which, respectively, produce mainly isothiocyanates, epithionitriles, and nitriles. In all three ecotypes, general stress-responsive genes, genes belonging to octadecanoid and indole glucosinolate synthesis pathways were induced upon both generalist and specialist attack. By contrast, transcription of myrosinases, enzymes hydrolysing glucosinolates, was suppressed. The induction of the jasmonic acid synthesis pathway was strongest in Cvi, while the up-regulation of the indole glucosinolate synthesis pathway was highest in Ler, suggesting a slightly different defence strategy in these two ecotypes. Specialist and generalist infestations caused statistically significant differential regulation of 60 genes in Ws and 21 in Cvi. Among these were jasmonic acid and tryptophan synthesis pathway enzymes, and pathogenesis related protein (PR1). Insect no-choice experiments revealed lowered fitness of B. brassicae on Ler and Cvi in comparison to Ws, but no ecotype-dependent change in fecundity of M. persicae. Targeted studies employing constructs of GUS reporter gene under the control of promoters from CYP79B2 and CYP79B3 genes showed insect-specific induction of the indole glucosinolates synthesis pathway.

Aetiology-specific patterns in end-stage heart failure patients identified by functional annotation and classification of microarray data.

BACKGROUND: The objective of the present study was to use gene expression profiling, functional annotations and classification to identify aetiology-specific biological processes and potential molecular markers for different aetiologies of end-stage heart failure. METHODS AND RESULTS: Individual left ventricular myocardial samples from eleven coronary artery disease and nine dilated cardiomyopathy transplant patients were co-hybridized with pooled RNA from four non-failing hearts on custom-made arrays of 7000 human genes. Significance analysis identified differential expression of 153 and 147 genes, respectively, in coronary artery disease or dilated cardiomyopathy versus non-failing hearts. Analysis of Gene Ontology biological process annotations indicated aetiology-specific patterns, primarily related to genes involved in catabolism and in regulation of protein kinase activity. Gene expression classifiers were obtained and used for class prediction of random samples of coronary artery diseased and dilated cardiomyopathic hearts. Best classifiers frequently included matrix metalloproteinase 3, fibulin 1, ATP-binding cassette, sub-family B member 1 and iroquois homeobox protein 5. CONCLUSION: Combining functional annotation from microarray data and classification analysis constitutes a potent strategy to identify disease-specific biological processes and gene expression markers in e.g. end-stage coronary artery disease and dilated cardiomyopathy.

Rough set-based proteochemometrics modeling of G-protein-coupled receptor-ligand interactions.

G-Protein-coupled receptors (GPCRs) are among the most important drug targets. Because of a shortage of 3D crystal structures, most of the drug design for GPCRs has been ligand-based. We propose a novel, rough set-based proteochemometric approach to the study of receptor and ligand recognition. The approach is validated on three datasets containing GPCRs. In proteochemometrics, properties of receptors and ligands are used in conjunction and modeled to predict binding affinity. The rough set (RS) rule-based models presented herein consist of minimal decision rules that associate properties of receptors and ligands with high or low binding affinity. The information provided by the rules is then used to develop a mechanistic interpretation of interactions between the ligands and receptors included in the datasets. The first two datasets contained descriptors of melanocortin receptors and peptide ligands. The third set contained descriptors of adrenergic receptors and ligands. All the rule models induced from these datasets have a high predictive quality. An example of a decision rule is "If R1_ligand(Ethyl) and TM helix 2 position 27(Methionine) then Binding(High)." The easily interpretable rule sets are able to identify determinative receptor and ligand parts. For instance, all three models suggest that transmembrane helix 2 is determinative for high and low binding affinity. RS models show that it is possible to use rule-based models to predict ligand-binding affinities. The models may be used to gain a deeper biological understanding of the combinatorial nature of receptor-ligand interactions.

Gene expression based classification of gastric carcinoma.

The aim of the present work is to identify molecular markers that allow classification of gastric carcinoma with respect to important clinicopathological parameters. Gastric adenocarcinomas were subjected to cDNA microarray analysis with a 2.504 gene probe set. Using the Rosetta rough-set based learning system, good classifiers were generated for gene-expression based prediction of intestinal or diffuse growth pattern according to Laurén's classification and presence of lymph node metastases. To our knowledge, this is the first study on gastric carcinoma in which molecular classification has been achieved for more than one clinicopathological parameter based on microarray gene expression profiles.

Predicting gene ontology biological process from temporal gene expression patterns.

The aim of the present study was to generate hypotheses on the involvement of uncharacterized genes in biological processes. To this end, supervised learning was used to analyze microarray-derived time-series gene expression data. Our method was objectively evaluated on known genes using cross-validation and provided high-precision Gene Ontology biological process classifications for 211 of the 213 uncharacterized genes in the data set used. In addition, new roles in biological process were hypothesized for known genes. Our method uses biological knowledge expressed by Gene Ontology and generates a rule model associating this knowledge with minimal characteristic features of temporal gene expression profiles. This model allows learning and classification of multiple biological process roles for each gene and can predict participation of genes in a biological process even though the genes of this class exhibit a wide variety of gene expression profiles including inverse coregulation. A considerable number of the hypothesized new roles for known genes were confirmed by literature search. In addition, many biological process roles hypothesized for uncharacterized genes were found to agree with assumptions based on homology information. To our knowledge, a gene classifier of similar scope and functionality has not been reported earlier.