A probabilistic model for the extraction of expression levels from oligonucleotide arrays

Transcriptomic analysis of the heat stress response for a commercial baker's yeast Saccharomyces cerevisiae

The aim of this study is to explore the effects of heat stresses on global gene expression profiles and to identify the candidate genes for the heat stress response in commercial baker's yeast (Saccharomyces cerevisiae) by using microarray technology and comparative statistical data analyses. The data from all hybridizations and array normalization were analyzed using the GeneSpringGX 12.1 (Agilent) and the R 2.15.2 program language. In the analysis, all required statistical methods were performed comparatively. For the normalization step, among alternatives, the RMA (Robust Microarray Analysis) results were used. To determine differentially expressed genes under heat stress treatments, the fold-change and the hypothesis testing approaches were executed under various cut-off values via different multiple testing procedures then the up/down regulated probes were functionally categorized via the PAMSAM clustering. The results of the analysis concluded that the transcriptome changes under the heat shock. Moreover, the temperature-shift stress treatments show that the number of differentially up-regulated genes among the heat shock proteins and transcription factors changed significantly. Finally, the change in temperature is one of the important environmental conditions affecting propagation and industrial application of baker's yeast. This study statistically analyzes this affect via one-channel microarray data.

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.

Genome-wide analyses of gene expression during mouse endochondral ossification

Background

Endochondral ossification is a complex process involving a series of events that are initiated by the establishment of a chondrogenic template and culminate in its replacement through the coordinated activity of osteoblasts, osteoclasts and endothelial cells. Comprehensive analyses of in vivo gene expression profiles during these processes are essential to obtain a complete understanding of the regulatory mechanisms involved.

Methodology/Principal Findings

To address these issues, we completed a microarray screen of three zones derived from manually segmented embryonic mouse tibiae. Classification of genes differentially expressed between each respective zone, functional categorization as well as characterization of gene expression patterns, cytogenetic loci, signaling pathways and functional motifs both confirmed reported data and provided novel insights into endochondral ossification. Parallel comparisons of the microdissected tibiae data set with our previously completed micromass culture screen further corroborated the suitability of micromass cultures for modeling gene expression in chondrocyte development. The micromass culture system demonstrated striking similarities to the in vivo microdissected tibiae screen; however, the micromass system was unable to accurately distinguish gene expression differences in the hypertrophic and mineralized zones of the tibia.

Conclusions/Significance

These studies allow us to better understand gene expression patterns in the growth plate and endochondral bones and provide an important technical resource for comparison of gene expression in diseased or experimentally-manipulated cartilages. Ultimately, this work will help to define the genomic context in which genes are expressed in long bones and to understand physiological and pathological ossification.

A comprehensive sensitivity analysis of microarray breast cancer classification under feature variability

BACKGROUND

Large discrepancies in signature composition and outcome concordance have been observed between different microarray breast cancer expression profiling studies. This is often ascribed to differences in array platform as well as biological variability. We conjecture that other reasons for the observed discrepancies are the measurement error associated with each feature and the choice of preprocessing method. Microarray data are known to be subject to technical variation and the confidence intervals around individual point estimates of expression levels can be wide. Furthermore, the estimated expression values also vary depending on the selected preprocessing scheme. In microarray breast cancer classification studies, however, these two forms of feature variability are almost always ignored and hence their exact role is unclear.

RESULTS

We have performed a comprehensive sensitivity analysis of microarray breast cancer classification under the two types of feature variability mentioned above. We used data from six state of the art preprocessing methods, using a compendium consisting of eight different datasets, involving 1131 hybridizations, containing data from both one and two-color array technology. For a wide range of classifiers, we performed a joint study on performance, concordance and stability. In the stability analysis we explicitly tested classifiers for their noise tolerance by using perturbed expression profiles that are based on uncertainty information directly related to the preprocessing methods. Our results indicate that signature composition is strongly influenced by feature variability, even if the array platform and the stratification of patient samples are identical. In addition, we show that there is often a high level of discordance between individual class assignments for signatures constructed on data coming from different preprocessing schemes, even if the actual signature composition is identical.

CONCLUSION

Feature variability can have a strong impact on breast cancer signature composition, as well as the classification of individual patient samples. We therefore strongly recommend that feature variability is considered in analyzing data from microarray breast cancer expression profiling experiments.

Genomic analysis of the function of the transcription factor gata3 during development of the mammalian inner ear

We have studied the function of the zinc finger transcription factor gata3 in auditory system development by analysing temporal profiles of gene expression during differentiation of conditionally immortal cell lines derived to model specific auditory cell types and developmental stages. We tested and applied a novel probabilistic method called the gamma Model for Oligonucleotide Signals to analyse hybridization signals from Affymetrix oligonucleotide arrays. Expression levels estimated by this method correlated closely (p<0.0001) across a 10-fold range with those measured by quantitative RT-PCR for a sample of 61 different genes. In an unbiased list of 26 genes whose temporal profiles clustered most closely with that of gata3 in all cell lines, 10 were linked to Insulin-like Growth Factor signalling, including the serine/threonine kinase Akt/PKB. Knock-down of gata3 in vitro was associated with a decrease in expression of genes linked to IGF-signalling, including IGF1, IGF2 and several IGF-binding proteins. It also led to a small decrease in protein levels of the serine-threonine kinase Akt2/PKBβ, a dramatic increase in Akt1/PKBα protein and relocation of Akt1/PKBα from the nucleus to the cytoplasm. The cyclin-dependent kinase inhibitor p27^kip1, a known target of PKB/Akt, simultaneously decreased. In heterozygous gata3 null mice the expression of gata3 correlated with high levels of activated Akt/PKB. This functional relationship could explain the diverse function of gata3 during development, the hearing loss associated with gata3 heterozygous null mice and the broader symptoms of human patients with Hearing-Deafness-Renal anomaly syndrome.

Construction and use of gene expression covariation matrix

BACKGROUND

One essential step in the massive analysis of transcriptomic profiles is the calculation of the correlation coefficient, a value used to select pairs of genes with similar or inverse transcriptional profiles across a large fraction of the biological conditions examined. Until now, the choice between the two available methods for calculating the coefficient has been dictated mainly by technological considerations. Specifically, in analyses based on double-channel techniques, researchers have been required to use covariation correlation, i.e. the correlation between gene expression changes measured between several pairs of biological conditions, expressed for example as fold-change. In contrast, in analyses of single-channel techniques scientists have been restricted to the use of coexpression correlation, i.e. correlation between gene expression levels. To our knowledge, nobody has ever examined the possible benefits of using covariation instead of coexpression in massive analyses of single channel microarray results.

RESULTS

We describe here how single-channel techniques can be treated like double-channel techniques and used to generate both gene expression changes and covariation measures. We also present a new method that allows the calculation of both positive and negative correlation coefficients between genes. First, we perform systematic comparisons between two given biological conditions and classify, for each comparison, genes as increased (I), decreased (D), or not changed (N). As a result, the original series of n gene expression level measures assigned to each gene is replaced by an ordered string of n(n-1)/2 symbols, e.g. IDDNNIDID....DNNNNNNID, with the length of the string corresponding to the number of comparisons. In a second step, positive and negative covariation matrices (CVM) are constructed by calculating statistically significant positive or negative correlation scores for any pair of genes by comparing their strings of symbols.

CONCLUSION

This new method, applied to four different large data sets, has allowed us to construct distinct covariation matrices with similar properties. We have also developed a technique to translate these covariation networks into graphical 3D representations and found that the local assignation of the probe sets was conserved across the four chip set models used which encompass three different species (humans, mice, and rats). The application of adapted clustering methods succeeded in delineating six conserved functional regions that we characterized using Gene Ontology information.

Microarray profiling reveals CXCR4a is downregulated by blood flow in vivo and mediates collateral formation in zebrafish embryos

The response to hemodynamic force is implicated in a number of pathologies including collateral vessel development. However, the transcriptional effect of hemodynamic force is extremely challenging to examine in vivo in mammals without also detecting confounding processes such as hypoxia and ischemia. We therefore serially examined the transcriptional effect of preventing cardiac contraction in zebrafish embryos which can be deprived of circulation without experiencing hypoxia since they obtain sufficient oxygenation by diffusion. Morpholino antisense knock-down of cardiac troponin T2 (tnnt2) prevented cardiac contraction without affecting vascular development. Gene expression in whole embryo RNA from tnnt2 or control morphants at 36, 48, and 60 h postfertilization (hpf) was assessed using Affymetrix GeneChip Zebrafish Genome Arrays (>14,900 transcripts). We identified 308 differentially expressed genes between tnnt2 and control morphants. One such (CXCR4a) was significantly more highly expressed in tnnt2 morphants at 48 and 60 hpf than controls. In situ hybridization localized CXCR4a upregulation to endothelium of both tnnt2 morphants and gridlock mutants (which have an occluded aorta preventing distal blood flow). This upregulation appears to be of functional significance as either CXCR4a knock-down or pharmacologic inhibition impaired the ability of gridlock mutants to recover blood flow via collateral vessels. We conclude absence of hemodynamic force induces endothelial CXCR4a upregulation that promotes recovery of blood flow.

Determining gene expression on a single pair of microarrays

Background

In microarray experiments the numbers of replicates are often limited due to factors such as cost, availability of sample or poor hybridization. There are currently few choices for the analysis of a pair of microarrays where N = 1 in each condition. In this paper, we demonstrate the effectiveness of a new algorithm called PINC (PINC is Not Cyber-T) that can analyze Affymetrix microarray experiments.

Results

PINC treats each pair of probes within a probeset as an independent measure of gene expression using the Bayesian framework of the Cyber-T algorithm and then assigns a corrected p-value for each gene comparison.

The p-values generated by PINC accurately control False Discovery rate on Affymetrix control data sets, but are small enough that family-wise error rates (such as the Holm's step down method) can be used as a conservative alternative to false discovery rate with little loss of sensitivity on control data sets.

Conclusion

PINC outperforms previously published methods for determining differentially expressed genes when comparing Affymetrix microarrays with N = 1 in each condition. When applied to biological samples, PINC can be used to assess the degree of variability observed among biological replicates in addition to analyzing isolated pairs of microarrays.

Propagating uncertainty in microarray data analysis

Microarray technology is associated with many sources of experimental uncertainty. In this review we discuss a number of approaches for dealing with this uncertainty in the processing of data from microarray experiments. We focus here on the analysis of high-density oligonucleotide arrays, such as the popular Affymetrix GeneChip array, which contain multiple probes for each target. This set of probes can be used to determine an estimate for the target concentration and can also be used to determine the experimental uncertainty associated with this measurement. This measurement uncertainty can then be propagated through the downstream analysis using probabilistic methods. We give examples showing how these credibility intervals can be used to help identify differential expression, to combine information from replicated experiments and to improve the performance of principal component analysis.

Expression profiling of Dexamethasone-treated primary chondrocytes identifies targets of glucocorticoid signalling in endochondral bone development

Background

Glucocorticoids (GCs) are widely used anti-inflammatory drugs. While useful in clinical practice, patients taking GCs often suffer from skeletal side effects including growth retardation in children and adolescents, and decreased bone quality in adults. On a physiological level, GCs have been implicated in the regulation of chondrogenesis and osteoblast differentiation, as well as maintaining homeostasis in cartilage and bone. We identified the glucocorticoid receptor (GR) as a potential regulator of chondrocyte hypertrophy in a microarray screen of primary limb bud mesenchyme micromass cultures. Some targets of GC regulation in chondrogenesis are known, but the global effects of pharmacological GC doses on chondrocyte gene expression have not been comprehensively evaluated.

Results

This study systematically identifies a spectrum of GC target genes in embryonic growth plate chondrocytes treated with a synthetic GR agonist, dexamethasone (DEX), at 6 and 24 hrs. Conventional analysis of this data set and gene set enrichment analysis (GSEA) was performed. Transcripts associated with metabolism were enriched in the DEX condition along with extracellular matrix genes. In contrast, a subset of growth factors and cytokines were negatively correlated with DEX treatment. Comparing DEX-induced gene expression data to developmental changes in gene expression in micromass cultures revealed an additional layer of complexity in which DEX maintains the expression of certain chondrocyte marker genes while inhibiting factors that promote vascularization and ultimately ossification of the cartilaginous template.

Conclusion

Together, these results provide insight into the mechanisms and major molecular classes functioning downstream of DEX in primary chondrocytes. In addition, comparison of our data with microarray studies of DEX treatment in other cell types demonstrated that the majority of DEX effects are tissue-specific. This study provides novel insights into the effects of pharmacological GC on chondrocyte gene transcription and establishes the foundation for subsequent functional studies.

FGX: a frequentist gene expression index for Affymetrix arrays

We consider a new frequentist gene expression index for Affymetrix oligonucleotide DNA arrays, using a similar probe intensity model as suggested by Hein and others (2005), called the Bayesian gene expression index (BGX). According to this model, the perfect match and mismatch values are assumed to be correlated as a result of sharing a common gene expression signal. Rather than a Bayesian approach, we develop a maximum likelihood algorithm for estimating the underlying common signal. In this way, estimation is explicit and much faster than the BGX implementation. The observed Fisher information matrix, rather than a posterior credibility interval, gives an idea of the accuracy of the estimators. We evaluate our method using benchmark spike-in data sets from Affymetrix and GeneLogic by analyzing the relationship between estimated signal and concentration, i.e. true signal, and compare our results with other commonly used methods.

Probe-level measurement error improves accuracy in detecting differential gene expression

MOTIVATION

Finding differentially expressed genes is a fundamental objective of a microarray experiment. Numerous methods have been proposed to perform this task. Existing methods are based on point estimates of gene expression level obtained from each microarray experiment. This approach discards potentially useful information about measurement error that can be obtained from an appropriate probe-level analysis. Probabilistic probe-level models can be used to measure gene expression and also provide a level of uncertainty in this measurement. This probe-level measurement error provides useful information which can help in the identification of differentially expressed genes.

RESULTS

We propose a Bayesian method to include probe-level measurement error into the detection of differentially expressed genes from replicated experiments. A variational approximation is used for efficient parameter estimation. We compare this approximation with MAP and MCMC parameter estimation in terms of computational efficiency and accuracy. The method is used to calculate the probability of positive log-ratio (PPLR) of expression levels between conditions. Using the measurements from a recently developed Affymetrix probe-level model, multi-mgMOS, we test PPLR on a spike-in dataset and a mouse time-course dataset. Results show that the inclusion of probe-level measurement error improves accuracy in detecting differential gene expression.

AVAILABILITY

The MAP approximation and variational inference described in this paper have been implemented in an R package pplr. The MCMC method is implemented in Matlab. Both software are available from http://umber.sbs.man.ac.uk/resources/puma.

Information for gene networks in inner ear development: a study centered on the transcription factor gata2

The search for molecular mechanisms to stimulate sensory regeneration in the mammalian inner ear is commonly based upon developmental studies. This has revealed many genes that regulate the differentiation of sensory cells. A major challenge is to place these genes into the context of functional networks that describe developmental processes more fully and increase the chances of identifying useful therapeutic targets. We used a novel approach to identify genes that are functionally related to the transcription factor gata2. Temporal profiles of gene expression were derived from three conditionally immortal cell lines and clustered to those of gata2 by applying the gamma model for oligonucleotide signals, a statistical method that allows quantitative analysis of oligonucleotide array data. We derived an objective list of 28 genes that clustered with gata2 in all three cell lines. A number of these genes have known functional links with gata2. Genes encoding CCAAT/enhancer binding proteins (C/EBP) and signal transducer and activation of transcription 3 (Stat3) are especially interesting as they are known to bind gata proteins directly. The results provide strong evidence that our experimental approach can reveal functional relationships between genes that regulate fundamental processes in the differentiation of sensory cells in the inner ear.

Comparison of Affymetrix GeneChip expression measures

MOTIVATION

In the Affymetrix GeneChip system, preprocessing occurs before one obtains expression level measurements. Because the number of competing preprocessing methods was large and growing we developed a benchmark to help users identify the best method for their application. A webtool was made available for developers to benchmark their procedures. At the time of writing over 50 methods had been submitted.

RESULTS

We benchmarked 31 probe set algorithms using a U95A dataset of spike in controls. Using this dataset, we found that background correction, one of the main steps in preprocessing, has the largest effect on performance. In particular, background correction appears to improve accuracy but, in general, worsen precision. The benchmark results put this balance in perspective. Furthermore, we have improved some of the original benchmark metrics to provide more detailed information regarding precision and accuracy. A handful of methods stand out as providing the best balance using spike-in data with the older U95A array, although different experiments on more current arrays may benchmark differently.

AVAILABILITY

The affycomp package, now version 1.5.2, continues to be available as part of the Bioconductor project (http://www.bioconductor.org). The webtool continues to be available at http://affycomp.biostat.jhsph.edu

CONTACT

rafa@jhu.edu

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Accounting for probe-level noise in principal component analysis of microarray data

MOTIVATION

Principal Component Analysis (PCA) is one of the most popular dimensionality reduction techniques for the analysis of high-dimensional datasets. However, in its standard form, it does not take into account any error measures associated with the data points beyond a standard spherical noise. This indiscriminate nature provides one of its main weaknesses when applied to biological data with inherently large variability, such as expression levels measured with microarrays. Methods now exist for extracting credibility intervals from the probe-level analysis of cDNA and oligonucleotide microarray experiments. These credibility intervals are gene and experiment specific, and can be propagated through an appropriate probabilistic downstream analysis.

RESULTS

We propose a new model-based approach to PCA that takes into account the variances associated with each gene in each experiment. We develop an efficient EM-algorithm to estimate the parameters of our new model. The model provides significantly better results than standard PCA, while remaining computationally reasonable. We show how the model can be used to 'denoise' a microarray dataset leading to improved expression profiles and tighter clustering across profiles. The probabilistic nature of the model means that the correct number of principal components is automatically obtained.

A tractable probabilistic model for Affymetrix probe-level analysis across multiple chips

MOTIVATION

Affymetrix GeneChip arrays are currently the most widely used microarray technology. Many summarization methods have been developed to provide gene expression levels from Affymetrix probe-level data. Most of the currently popular methods do not provide a measure of uncertainty for the expression level of each gene. The use of probabilistic models can overcome this limitation. A full hierarchical Bayesian approach requires the use of computationally intensive MCMC methods that are impractical for large datasets. An alternative computationally efficient probabilistic model, mgMOS, uses Gamma distributions to model specific and non-specific binding with a latent variable to capture variations in probe affinity. Although promising, the main limitations of this model are that it does not use information from multiple chips and does not account for specific binding to the mismatch (MM) probes.

RESULTS

We extend mgMOS to model the binding affinity of probe-pairs across multiple chips and to capture the effect of specific binding to MM probes. The new model, multi-mgMOS, provides improved accuracy, as demonstrated on some bench-mark datasets and a real time-course dataset, and is much more computationally efficient than a competing hierarchical Bayesian approach that requires MCMC sampling. We demonstrate how the probabilistic model can be used to estimate credibility intervals for expression levels and their log-ratios between conditions.

AVAILABILITY

Both mgMOS and the new model multi-mgMOS have been implemented in an R package, which is available at http://www.bioinf.man.ac.uk/resources/puma.

Ventral otic cell lines as developmental models of auditory epithelial and neural precursors

Conditionally immortal cell lines were established from the ventral otocyst of the Immortomouse at embryonic day 10.5 and selected to represent precursors of auditory sensory neural and epithelial cells. Selection was based upon dissection, tissue-specific markers, and expression of the transcription factor GATA3. Two cell lines expressed GATA3 but possessed intrinsically different genetic programs under differentiating conditions. US/VOT-E36 represented epithelial progenitors with potential to differentiate into sensory and nonsensory epithelial cells. US/VOT-N33 represented migrating neuroblasts. Under differentiating conditions in vitro the cell lines expressed very different gene expression profiles. Expression of several cell- and tissue-specific markers, including the transcription factors Pax2, GATA3, and NeuroD, differed between the cell lines in a pattern consistent with that observed between their counterparts in vivo. We suggest that these and other conditionally immortal cell lines can be used to study transient events in development against different backgrounds of cell competence.