Assessing differential gene expression with small sample sizes in oligonucleotide arrays using a mean-variance model

The identification of the genes that are differentially expressed in two-sample microarray experiments remains a difficult problem when the number of arrays is very small. We discuss the implications of using ordinary t-statistics and examine other commonly used variants. For oligonucleotide arrays with multiple probes per gene, we introduce a simple model relating the mean and variance of expression, possibly with gene-specific random effects. Parameter estimates from the model have natural shrinkage properties that guard against inappropriately small variance estimates, and the model is used to obtain a differential expression statistic. A limiting value to the positive false discovery rate (pFDR) for ordinary t-tests provides motivation for our use of the data structure to improve variance estimates. Our approach performs well compared to other proposed approaches in terms of the false discovery rate.     

Efficient oligonucleotide probe selection for pan-genomic tiling arrays

Background  

Array comparative genomic hybridization is a fast and cost-effective method for detecting, genotyping, and comparing the genomic sequence of unknown bacterial isolates. This method, as with all microarray applications, requires adequate coverage of probes targeting the regions of interest. An unbiased tiling of probes across the entire length of the genome is the most flexible design approach. However, such a whole-genome tiling requires that the genome sequence is known in advance. For the accurate analysis of uncharacterized bacteria, an array must query a fully representative set of sequences from the species' pan-genome. Prior microarrays have included only a single strain per array or the conserved sequences of gene families. These arrays omit potentially important genes and sequence variants from the pan-genome.     

A method for cross-species gene expression analysis with high-density oligonucleotide arrays

DNA microarrays have been widely used in gene expression analysis of biological processes. Due to a lack of sequence information, the applications have been largely restricted to humans and a few model organisms. Presented within this study are results of the cross-species hybridization with Affymetrix human high-density oligonucleotide arrays or GeneChip® using distantly related mammalian species; cattle, pig and dog. Based on the unique feature of the Affymetrix GeneChip® where every gene is represented by multiple probes, we hypothesized that sequence conservation within mammals is high enough to generate sufficient signals from some of the probes for expression analysis. We demonstrated that while overall hybridization signals are low for cross-species hybridization, a few probes of most genes still generated signals equivalent to the same-species hybridization. By masking the poorly hybridized probes electronically, the remaining probes provided reliable data for gene expression analysis. We developed an algorithm to select the reliable probes for analysis utilizing the match/mismatch feature of GeneChip®. When comparing gene expression between two tissues using the selected probes, we found a linear correlation between the cross-species and same-species hybridization. In addition, we validated cross-species hybridization results by quantitative PCR using randomly selected genes. The method shown herein could be applied to both plant and animal research.     

Hybridization modeling of oligonucleotide SNP arrays for accurate DNA copy number estimation

Affymetrix SNP arrays have been widely used for single-nucleotide polymorphism (SNP) genotype calling and DNA copy number variation inference. Although numerous methods have achieved high accuracy in these fields, most studies have paid little attention to the modeling of hybridization of probes to off-target allele sequences, which can affect the accuracy greatly. In this study, we address this issue and demonstrate that hybridization with mismatch nucleotides (HWMMN) occurs in all SNP probe-sets and has a critical effect on the estimation of allelic concentrations (ACs). We study sequence binding through binding free energy and then binding affinity, and develop a probe intensity composite representation (PICR) model. The PICR model allows the estimation of ACs at a given SNP through statistical regression. Furthermore, we demonstrate with cell-line data of known true copy numbers that the PICR model can achieve reasonable accuracy in copy number estimation at a single SNP locus, by using the ratio of the estimated AC of each sample to that of the reference sample, and can reveal subtle genotype structure of SNPs at abnormal loci. We also demonstrate with HapMap data that the PICR model yields accurate SNP genotype calls consistently across samples, laboratories and even across array platforms.     

Long oligonucleotide arrays on nylon for large-scale gene expression analysis

To date, most studies of multigenic expression patterns by long DNA array have used DNA fragments as probes. These probes are usually obtained as PCR products, and this represents a time-consuming and error-prone approach, requiring strict quality control. The present study examines the use of 40- and 70-mer synthetic oligonucleotides as probes for DNA array analysis with radioactive labeled targets. Design, spotting onto nylon filters, and hybridization conditions were determined and optimized. In this approach, the sensitivity and the specificity of the hybridization appear comparable to the conventional long DNA probes assay, permitting the analysis of small samples of approximately 1 microg total RNA. The long oligonucleotide array thus provides a very convenient method for the analysis of gene expression patterns in biological specimens and in clinical research.     

Theoretical and experimental comparisons of gene expression indexes for oligonucleotide arrays

MOTIVATION: Oligonucleotide expression arrays exhibit systematic and reproducible variation produced by the multiple distinct probes used to represent a gene. Recently, a gene expression index has been proposed that explicitly models probe effects, and provides improved fits of hybridization intensity for arrays containing perfect match (PM) and mismatch (MM) probe pairs. RESULTS: Here we use a combination of analytical arguments and empirical data to show directly that the estimates provided by model-based expression indexes are superior to those provided by commercial software. The improvement is greatest for genes in which probe effects vary substantially, and modeling the PM and MM intensities separately is superior to using the PM-MM differences. To empirically compare expression indexes, we designed a mixing experiment involving three groups of human fibroblast cells (serum starved, serum stimulated, and a 50:50 mixture of starved/stimulated), with six replicate HuGeneFL arrays in each group. Careful spiking of control genes provides evidence that 88-98% of the genes on the array are detectably transcribed, and that the model-based estimates can accurately detect the presence versus absence of a gene. The use of extensive replication from single RNA sources enables exploration of the technical variability of the array.     

A high performance test of differential gene expression for oligonucleotide arrays

Logit-t employs a logit-transformation for normalization followed by statistical testing at the probe-level. Using four publicly-available datasets, together providing 2,710 known positive incidences of differential expression and 2,913,813 known negative incidences, performance of statistical tests were: Logit-t provided 75% positive-predictive value, compared with 5% for Affymetrix Microarray Suite 5, 6% for dChip perfect match (PM)-only, and 9% for Robust Multi-array Analysis at the p < 0.01 threshold. Logit-t provided 70% sensitivity, Microarray Suite 5 provided 46%, dChip provided 53% and Robust Multi-array Analysis provided 63%.     

Evaluating machine learning approaches for aiding probe selection for gene-expression arrays

MOTIVATION: Microarrays are a fast and cost-effective method of performing thousands of DNA hybridization experiments simultaneously. DNA probes are typically used to measure the expression level of specific genes. Because probes greatly vary in the quality of their hybridizations, choosing good probes is a difficult task. If one could accurately choose probes that are likely to hybridize well, then fewer probes would be needed to represent each gene in a gene-expression microarray, and, hence, more genes could be placed on an array of a given physical size. Our goal is to empirically evaluate how successfully three standard machine-learning algorithms-na?ve Bayes, decision trees, and artificial neural networks-can be applied to the task of predicting good probes. Fortunately it is relatively easy to get training examples for such a learning task: place various probes on a gene chip, add a sample where the corresponding genes are highly expressed, and then record how well each probe measures the presence of its corresponding gene. With such training examples, it is possible that an accurate predictor of probe quality can be learned. RESULTS: Two of the learning algorithms we investigate-na?ve Bayes and neural networks-learn to predict probe quality surprisingly well. For example, in the top ten predicted probes for a given gene not used for training, on average about five rank in the top 2.5% of that gene's hundreds of possible probes. Decision-tree induction and the simple approach of using predicted melting temperature to rank probes perform significantly worse than these two algorithms. The features we use to represent probes are very easily computed and the time taken to score each candidate probe after training is minor. Training the na?ve Bayes algorithm takes very little time, and while it takes over 10 times as long to train a neural network, that time is still not very substantial (on the order of a few hours on a desktop workstation). We also report the information contained in the features we use to describe the probes. We find the fraction of cytosine in the probe to be the most informative feature. We also find, not surprisingly, that the nucleotides in the middle of the probes sequence are more informative than those at the ends of the sequence.     

Selecting oligonucleotide probes for whole-genome tiling arrays with a cross-hybridization potential

For designing oligonucleotide tiling arrays popular, current methods still rely on simple criteria like Hamming distance or longest common factors, neglecting base stacking effects which strongly contribute to binding energies. Consequently, probes are often prone to cross-hybridization which reduces the signal-to-noise ratio and complicates downstream analysis. We propose the first computationally efficient method using hybridization energy to identify specific oligonucleotide probes. Our Cross-Hybridization Potential (CHP) is computed with a Nearest Neighbor Alignment, which efficiently estimates a lower bound for the Gibbs free energy of the duplex formed by two DNA sequences of bounded length. It is derived from our simplified reformulation of t-gap insertion-deletion-like metrics. The computations are accelerated by a filter using weighted ungapped q-grams to arrive at seeds. The computation of the CHP is implemented in our software OSProbes, available under the GPL, which computes sets of viable probe candidates. The user can choose a trade-off between running time and quality of probes selected. We obtain very favorable results in comparison with prior approaches with respect to specificity and sensitivity for cross-hybridization and genome coverage with high-specificity probes. The combination of OSProbes and our Tileomatic method, which computes optimal tiling paths from candidate sets, yields globally optimal tiling arrays, balancing probe distance, hybridization conditions, and uniqueness of hybridization.     

Leveraging two-way probe-level block design for identifying differential gene expression with high-density oligonucleotide arrays

Background  

To identify differentially expressed genes across experimental conditions in oligonucleotide microarray experiments, existing statistical methods commonly use a summary of probe-level expression data for each probe set and compare replicates of these values across conditions using a form of the t-test or rank sum test. Here we propose the use of a statistical method that takes advantage of the built-in redundancy architecture of high-density oligonucleotide arrays.     

Probabilistic analysis of probe reliability in differential gene expression studies with short oligonucleotide arrays

Probe defects are a major source of noise in gene expression studies. While existing approaches detect noisy probes based on external information such as genomic alignments, we introduce and validate a targeted probabilistic method for analyzing probe reliability directly from expression data and independently of the noise source. This provides insights into the various sources of probe-level noise and gives tools to guide probe design.     

Combining gene expression data from different generations of oligonucleotide arrays

Background  

One of the important challenges in microarray analysis is to take full advantage of previously accumulated data, both from one's own laboratory and from public repositories. Through a comparative analysis on a variety of datasets, a more comprehensive view of the underlying mechanism or structure can be obtained. However, as we discover in this work, continual changes in genomic sequence annotations and probe design criteria make it difficult to compare gene expression data even from different generations of the same microarray platform.     

Genetic diversity in yeast assessed with whole-genome oligonucleotide arrays

The availability of a complete genome sequence allows the detailed study of intraspecies variability. Here we use high-density oligonucleotide arrays to discover 11,115 single-feature polymorphisms (SFPs) existing in one or more of 14 different yeast strains. We use these SFPs to define regions of genetic identity between common laboratory strains of yeast. We assess the genome-wide distribution of genetic variation on the basis of this yeast population. We find that genome variability is biased toward the ends of chromosomes and is more likely to be found in genes with roles in fermentation or in transport. This subtelomeric bias may arise through recombination between nonhomologous sequences because full-gene deletions are more common in these regions than in more central regions of the chromosome.     

Global assessment of cross-hybridization for oligonucleotide arrays.

Cross-hybridization is the tendency for chains of nucleic acids to bind to other chains of nucleic acids that have similar but not identical sequences. This has the potential to make the interpretation of microarray experiments difficult since intensity at a spot on the array does not simply depend on the quantity of target in the sample. We propose a method for evaluating the extent of cross-hybridization in oligonucleotide arrays for data arising from a typical microarray experiment. We show that the level of cross-hybridization can be quite substantial. We argue that this makes the interpretation of the difference calls provided by MAS 5.0 difficult.     

Meta-analysis of gene expression microarrays with missing replicates

Background  

Many different microarray experiments are publicly available today. It is natural to ask whether different experiments for the same phenotypic conditions can be combined using meta-analysis, in order to increase the overall sample size. However, some genes are not measured in all experiments, hence they cannot be included or their statistical significance cannot be appropriately estimated in traditional meta-analysis. Nonetheless, these genes, which we refer to as incomplete genes, may also be informative and useful.