Statistical approaches in the analysis of gene expression data derived from bone regeneration specific cDNA microarrays

Recent advances in molecular biology (e.g., cDNA microarray technology) enables the simultaneous monitoring of the expression level of thousands of genes. Due to the massive amount of complex data generated, sophisticated statistical approaches are necessary in order to properly address the experimental investigation. In this paper, we present statistical analysis of cDNA microarray data derived from bone regeneration experiments. Several interesting features from these data distinguish it from commonly used microarray experiment (i.e., separate hybridization of mRNA samples from reference and experimental tissues, selectively spotted cDNA sequences and 1060 systematically selected blank spots included in each array). Using this data set, we propose new methods for bioinformatic data normalization, as well as the modification and application of various other published methods in order to identify co-regulated gene expression patterns during the healing of a bone fracture. The proposed normalization methods perform effectively to eliminate the variations with a simple algorithm. Results from our cluster analysis revealed several clusters having distinct gene expression patterns during fracture healing. Our simulation study supports the reliability of the proposed methods.     

Systematic Approaches for Incorporating Control Spots and Data Quality Information to Improve Normalization of cDNA Microarray Data

Background: Normalization and data quality control are two important aspects in microarray data analysis. Proper normalization and data quality control ensure that intensity ratios provide meaningful and accurate measurement of relative gene expression values. Control spots such as spikes and housekeeping genes with known concentrations in two channels are often used for calibrating experimental parameters. They provide valuable information about experimental variation which can be utilized for better normalization. They are also needed for proper normalization in cases that the most of the spots tend to change in one direction. In addition, it is desirable to include information on spot quality. Such information is available in a typical microarray data set, but is not fully utilized by existing normalization methods.

Results: We propose two extensions of the two-way semi-linear model (TW-SLM) for appropriately combining control genes and spot quality information in normalization. The first extension (TW-SLMC) is designed to systematically incorporate control spots in a semi-parametric model to calibrate estimated normalization curves so that the relative fold changes of gene expressions are accurately estimated. Extrapolation is not required in this approach. The second extension (TW-SLMQ) is proposed to incorporate spot quality measure into normalization. This approach down-weights spots with lower quality scores in normalization. These two extensions can be used simultaneously for normalizing a data set. Two microarray data sets are used to demonstrate the proposed methods. Availability: An R based computing package is developed for the proposed methods and available from the corresponding authors.

Contact: Deli Wang: deliwang@uab.edu or Jian Huang: jian-huang@uiowa.edu.     

Empirical Bayes Gene Screening Tool for Time-Course or Dose–Response Microarray Data

Abstract

An efficient method to reduce the dimensionality of microarray gene expression data from thousands or tens of thousands of cDNA clones down to a subset of the most differentially expressed cDNA clones is essential in order to simplify the massive amount of data generated from microarray experiments. An extension to the methods of Efron et al. [Efron, B., Tibshirani, R., Storey, J., Tusher, V. (2001). Empirical Bayes analysis of a microarray experiment. J. Am. Statist. Assoc. 96:1151–1160] is applied to a differential time-course experiment to determine a subset of cDNAs that have the largest probability of being differentially expressed with respect to treatment conditions across a set of unequally spaced time points. The proposed extension, which is advocated to be a screening tool, allows for inference across a continuous variable in addition to incorporating a more complex experimental design and allowing for multiple design replications. With the current data the focus is on a time-course experiment; however, the proposed methods can easily be implemented on a dose–response experiment, or any other microarray experiment that contains a continuous variable of interest. The proposed empirical Bayes gene-screening tool is compared with the Efron et al. (2001) method in addition to an adjusted model-based t-value using a time-course data set where the toxicological effect of a specific mixture of chemicals is being studied.     

Empirical bayes gene screening tool for time-course or dose-response microarray data

An efficient method to reduce the dimensionality of microarray gene expression data from thousands or tens of thousands of cDNA clones down to a subset of the most differentially expressed cDNA clones is essential in order to simplify the massive amount of data generated from microarray experiments. An extension to the methods of Efron et al. [Efron, B., Tibshirani, R., Storey, J., Tusher, V. (2001). Empirical Bayes analysis of a microarray experiment. J. Am. Statist. Assoc. 96:1151-1160] is applied to a differential time-course experiment to determine a subset of cDNAs that have the largest probability of being differentially expressed with respect to treatment conditions across a set of unequally spaced time points. The proposed extension, which is advocated to be a screening tool, allows for inference across a continuous variable in addition to incorporating a more complex experimental design and allowing for multiple design replications. With the current data the focus is on a time-course experiment; however, the proposed methods can easily be implemented on a dose-response experiment, or any other microarray experiment that contains a continuous variable of interest. The proposed empirical Bayes gene-screening tool is compared with the Efron et al. (2001) method in addition to an adjusted model-based t-value using a time-course data set where the toxicological effect of a specific mixture of chemicals is being studied.     

Systematic approaches for incorporating control spots and data quality information to improve normalization of cDNA microarray data

BACKGROUND: Normalization and data quality control are two important aspects in microarray data analysis. Proper normalization and data quality control ensure that intensity ratios provide meaningful and accurate measurement of relative gene expression values. Control spots such as spikes and housekeeping genes with known concentrations in two channels are often used for calibrating experimental parameters. They provide valuable information about experimental variation which can be utilized for better normalization. They are also needed for proper normalization in cases that the most of the spots tend to change in one direction. In addition, it is desirable to include information on spot quality. Such information is available in a typical microarray data set, but is not fully utilized by existing normalization methods. RESULTS: We propose two extensions of the two-way semi-linear model (TW-SLM) for appropriately combining control genes and spot quality information in normalization. The first extension (TW-SLMC) is designed to systematically incorporate control spots in a semi-parametric model to calibrate estimated normalization curves so that the relative fold changes of gene expressions are accurately estimated. Extrapolation is not required in this approach. The second extension (TW-SLMQ) is proposed to incorporate spot quality measure into normalization. This approach down-weights spots with lower quality scores in normalization. These two extensions can be used simultaneously for normalizing a data set. Two microarray data sets are used to demonstrate the proposed methods. Availability: An R based computing package is developed for the proposed methods and available from the corresponding authors.     

A generalized additive model for microarray gene expression data analysis

Microarray technology allows the measurement of expression levels of a large number of genes simultaneously. There are inherent biases in microarray data generated from an experiment. Various statistical methods have been proposed for data normalization and data analysis. This paper proposes a generalized additive model for the analysis of gene expression data. This model consists of two sub-models: a non-linear model and a linear model. We propose a two-step normalization algorithm to fit the two sub-models sequentially. The first step involves a non-parametric regression using lowess fits to adjust for non-linear systematic biases. The second step uses a linear ANOVA model to estimate the remaining effects including the interaction effect of genes and treatments, the effect of interest in a study. The proposed model is a generalization of the ANOVA model for microarray data analysis. We show correspondences between the lowess fit and the ANOVA model methods. The normalization procedure does not assume the majority of genes do not change their expression levels, and neither does it assume two channel intensities from the same spot are independent. The procedure can be applied to either one channel or two channel data from the experiments with multiple treatments or multiple nuisance factors. Two toxicogenomic experiment data sets and a simulated data set are used to contrast the proposed method with the commonly known lowess fit and ANOVA methods.     

Identification of Differentially Expressed Genes with Multivariate Outlier Analysis

Abstract

DNA microarray offers a powerful and effective technology to monitor the changes in the gene expression levels for thousands of genes simultaneously. It is being widely applied to explore the quantitative alternation in gene regulation in response to a variety of aspects including diseases and exposure of toxicant. A common task in analyzing microarray data is to identify the differentially expressed genes under two different experimental conditions. Because of the large number of genes and small number of arrays, and higher signal-noise ratio in microarray data, many traditional approaches seem improper. In this paper, a multivariate mixture model is applied to model the expression level of replicated arrays, considering the differentially expressed genes as the outliers of the expression data. In order to detect the outliers of the multivariate mixture model, an effective and robust statistical method is first applied to microarray analysis. This method is based on the analysis of kurtosis coefficient (KC) of the projected multivariate data arising from a mixture model so as to identify the outliers. We utilize the multivariate KC algorithm to our microarray experiment with the control and toxic treatment. After the processing of data, the differential genes are successfully identified from 1824 genes on the UCLA M07 microarray chip. We also use the RT-PCR method and two robust statistical methods, minimum covariance determinant (MCD) and minimum volume ellipsoid (MVE), to verify the expression level of outlier genes identified by KC algorithm. We conclude that the robust multivariate tool is practical and effective for the detection of differentially expressed genes.     

Treatment of microarray experiments as split-plot designs

This paper shows that microarray experiments are split-plot, or split-unit, designs. The larger size experimental unit (the whole plot) is the array, and the treatment applied to this unit is the treatment given to the cells which produce the cDNA that is hybridized to the array. The smaller size experimental unit (the subplot) is the spot on the array, and the treatment applied to this unit is the gene giving rise to the DNA or oligonucleotide attached at that spot. Various treatment and design structures can be applied to the whole plot and the subplot; we consider the model equations appropriate to different designs. Preliminary normalization of the data can be avoided by including appropriate blocking terms in the model equation. We show how conventional analysis of variance can be used to test for significant differences in expression, and consider multiplicity corrections and graphical methods for identifying important expression differences.     

Sclerostin monoclonal antibodies on bone metabolism and fracture healing

Introduction: The biological enhancement of fracture healing may prevent complications such as non-union and revision surgery. Sclerostin is produced by osteocytes and binds to the LRP5/6 receptor. This inhibits the Wnt signalling pathway and thereby reduces bone formation.

Areas covered: Targeted deletion of the sclerostin gene has been found to enhance bone formation and fracture healing in rodent models. A number of in vivo studies have investigated the effect of sclerostin antibody on bone density with promising results. It also has an ability to promote fracture healing, screw fixation and metaphyseal bone healing in vivo. Early clinical studies have also demonstrated that it can increase bone mineral density, whilst being safe and well tolerated by patients.

Expert opinion: The data support the further investigation of this agent for the promotion of fracture healing. We aim to review the current literature and present an update on the use of this agent to promote bone formation and healing.     

应用cDNA微阵列对MDS骨髓细胞基因表达谱的初步研究

目的应用cDNA微阵列初步研究骨髓增生异常综合征(MDS)的基因表达谱。方法将2例患骨髓单个核细胞RNA各取等量混合后进行逆转录Cy5标记，和Cy3标记的正常对照cDNA混合，然后与H141微阵列杂交。结果H141s芯片中共点样13484个基因克隆，其中1064个靶克隆是针对同一基因内不同序列的cDNA片段，重复点样至少2次，表达结果在2张芯片内各自完全一致的分别为625个(58．7％)和630个(59．2％)，而2张芯片之间对应位置点检测结果完全相同的为783个(73．6％)。MDS患存在表达差异的基因为409个，其中101个基因参与造血调控，主要涉及转录因子、细胞周期调节蛋白、代谢相关基因、表面黏附分子等。结论cDNA微阵列可高效提供丰富的基因表达信息，并为深入研究MDS的发病分子机理提供线索。重复点样实验可降低微阵列操作过程引起的表达偏差。     

Statistical Methods for Analyzing Tissue Microarray Data

Abstract

Tissue microarrays (TMAs) are a new high-throughput tool for the study of protein expression patterns in tissues and are increasingly used to evaluate the diagnostic and prognostic importance of biomarkers. TMA data are rather challenging to analyze. Covariates are highly skewed, non-normal, and may be highly correlated. We present statistical methods for relating TMA data to censored time-to-event data. We review methods for evaluating the predictive power of Cox regression models and show how to test whether biomarker data contain predictive information above and beyond standard pathology covariates. We use nonparametric bootstrap methods to validate model fitting indices such as the concordance index. We also present data mining methods for characterizing high risk patients with simple biomarker rules. Since researchers in the TMA community routinely dichotomize biomarker expression values, survival trees are a natural choice. We also use bump hunting (patient rule induction method), which we adapt to the use with survival data. The proposed methods are applied to a kidney cancer tissue microarray data set.     

Assessing the variability in GeneChip data

Introduction: Oligonucleotide and cDNA microarray experiments are now common practice in biological science research. The goal of these experiments is generally to gain clues about the functions of genes by measuring how their expression levels rise and fall in response to changing experimental conditions. Measures of gene expression are affected, however, by a variety of factors. This paper introduces statistical methods to assess the variability of Affymetrix GeneChip® data due to randomness. Methods: The variation of Affymetrix’s GeneChip® signal data are quantified at both chip level and individual gene level, respectively, by the agreement study method and variance components method. Three agreement measurement methods are introduced to assess the variability among chips. Variation sources for gene expression data are decomposed into four categories: systematic experiment variation, treatment effect, biological variation, and chip variation. The focus of this paper is on evaluating and comparing the last two kinds of variations. Results: Measurement of agreement and variance components methods were applied to an experimental data, and the calculation and interpretation were exemplified. The variability between biological samples were shown to exist and were assessed at both the chip level and individual gene level. Using the variance components method, it was found that the biological and chip variation are roughly comparable. The Statistical Analysis System (SAS) program for doing the agreement studies can be obtained from the correspondence author.     

Identification of differentially expressed genes with multivariate outlier analysis

DNA microarray offers a powerful and effective technology to monitor the changes in the gene expression levels for thousands of genes simultaneously. It is being widely applied to explore the quantitative alternation in gene regulation in response to a variety of aspects including diseases and exposure of toxicant. A common task in analyzing microarray data is to identify the differentially expressed genes under two different experimental conditions. Because of the large number of genes and small number of arrays, and higher signal-noise ratio in microarray data, many traditional approaches seem improper. In this paper, a multivariate mixture model is applied to model the expression level of replicated arrays, considering the differentially expressed genes as the outliers of the expression data. In order to detect the outliers of the multivariate mixture model, an effective and robust statistical method is first applied to microarray analysis. This method is based on the analysis of kurtosis coefficient (KC) of the projected multivariate data arising from a mixture model so as to identify the outliers. We utilize the multivariate KC algorithm to our microarray experiment with the control and toxic treatment. After the processing of data, the differential genes are successfully identified from 1824 genes on the UCLA M07 microarray chip. We also use the RT-PCR method and two robust statistical methods, minimum covariance determinant (MCD) and minimum volume ellipsoid (MVE), to verify the expression level of outlier genes identified by KC algorithm. We conclude that the robust multivariate tool is practical and effective for the detection of differentially expressed genes.     

Non-steroidal anti-inflammatory drugs, cyclooxygenase-2 and the bone healing process

Traditional non-steroidal anti-inflammatory drugs (NSAID) and selective cyclooxygenase-2 (COX-2) inhibitors are widely used in the treatment of pain, including bone fracture pain and orthopaedic post-operative pain. The gastrointestinal and cardiovascular adverse effects of NSAIDs are acknowledged, but their effects on bone are less widely known. Prostaglandins play an important role in the regulation of osteoblast and osteoclast functions, and inhibition of prostaglandin production retards bone formation. Therefore, NSAIDs could be expected to have significant consequences in divergent clinical situations where bone formation or remodelling is a contributing factor. The present survey reviews current experimental and clinical evidence related to two of those conditions (i.e. on ectopic bone formation and on bone fracture healing). NSAIDs are used clinically to prevent ectopic bone formation (also known as heterotopic ossification) (e.g. after total hip arthroplasty or trauma). The efficacy of NSAIDs in the avoidance of heterotopic ossification has been documented in controlled clinical trials, but the inherent risks (e.g. on healing processes and on loosening of prostheses) need further studies. At the same time, NSAIDs are widely used in the treatment of fracture pain, and their inhibitory effects on the ongoing bone healing process have raised concerns. Results of fracture healing studies in animals treated with NSAIDs or in mice lacking COX-2 gene show that inhibition or deficiency of COX-2 impairs the bone healing process. The limited clinical data also support the assumption that inhibition of COX-2 by non-selective or COX-2-selective NSAIDs delays fracture healing. However, the clinical significance of the effect in various patient groups needs to be carefully assessed and further investigations are needed to characterize the patients at the highest risk for NSAID-induced delayed fracture healing and its complications. In the meantime, use of NSAIDs in fracture patients should be cautious, keeping in mind the benefits of pain relief and inhibition of ectopic bone formation on one hand, and the risks of non-union and retarded union on the other hand.     

Healing of the bone with anti-fracture drugs

Introduction: Fracture healing is a complex process that leads to the restoration of tissue integrity through bone repair and represents a unique physiological characteristic of bone. Developing a better understanding of a fracture is essential to plan best noninvasive treatment for the patient. In osteoporosis, the patient who suffers of a fragility fracture is recommended to initiate a treatment with compounds active in preventing other low-energy skeletal trauma. Pharmaceutical industries are developing controlled clinical trials aiming to evaluate the capability of osteoporosis drugs to accelerate fracture healing.

Areas covered: In preparing this review, a search was made with key words encompassing ‘osteoporosis anti-fracture drugs and bone repair/healing', ‘antiresorptives and bone repair/healing', ‘bone-forming agents and bone repair/healing', and ‘osteoporosis/anti-fracture drugs in fractures'. The results published in the area of the use of registered anti-fracture drugs to improve fracture repair and the efforts made to recommend measures for clinical outcomes in fracture healing acceleration are described in this report.

Expert opinion: At present, the use of systemic pharmacological agents active to improve fracture healing by the clinicians is controversial and clinicians and scientists must do a better job in determining the methods of assessment for fracture healing.     

Temporal and dose-dependent hepatic gene expression patterns in mice provide new insights into TCDD-Mediated hepatotoxicity.

In an effort to further characterize the mechanisms of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)-mediated toxicity, comprehensive temporal and dose-response microarray analyses were performed on hepatic tissue from immature ovariectomized C57BL/6 mice treated with TCDD. For temporal analysis, mice were gavaged with 30 microg/kg of TCDD or vehicle and sacrificed after 2, 4, 8, 12, 18, 24, 72, or 168 h. Dose-response mice were gavaged with 0, 0.001, 0.01, 0.1, 1, 10, 100, or 300 microg/kg of TCDD and sacrificed after 24 h. Hepatic gene expression profiles were monitored using custom cDNA microarrays containing 13,362 cDNA clones. Gene expression analysis identified 443 and 315 features which exhibited a significant change at one or more doses or time points, respectively, as determined using an empirical Bayes approach. Functional gene annotation extracted from public databases associated gene expression changes with physiological processes such as oxidative stress and metabolism, differentiation, apoptosis, gluconeogenesis, and fatty acid uptake and metabolism. Complementary histopathology (H&E and Oil Red O stains), clinical chemistry (i.e., alanine aminotransferase [ALT], triglyceride [TG], free fatty acids [FFA], cholesterol) and high-resolution gas chromatography/mass spectrometry assessment of hepatic TCDD levels were also performed in order to phenotypically anchor changes in gene expression to physiological end points. Collectively, the data support a proposed mechanism for TCDD-mediated hepatotoxicity, including fatty liver, which involves mobilization of peripheral fat and inappropriate increases in hepatic uptake of fatty acids.     

Investigational drugs for fracture healing: preclinical & clinical data

Introduction: The need for fracture healing enhancement for the management of fracture complications such as non-union and for the achievement of early function in fracture patients is constantly increasing. Therefore, the development and evaluation of novel pharmaceutical agents is mandatory in order to accelerate the process and increase bone union rates.

Areas covered: This review summarizes the most recent knowledge on the pharmacological enhancement of fracture repair. It provides a synopsis of the most important preclinical and clinical studies published over the past five years on long bone fracture healing.

Expert opinion: To date, limited drugs seem to have the potential for clinical use in fracture healing enhancement and the field is progressing very slowly. Among anti-osteoporotic drugs, only PTH and anti-sclerostin antibodies have such a potential but further research is needed before clinical use. The same applies also to BMPs, the use of which still carries major drawbacks that should be overcome before their widespread clinical utilization. Other drugs and growth factors, such as statins, VEGF, FGF, EPO, could be future key players in fracture healing but evidence is still lacking. Further in depth understanding of the healing process is essential in order to identify novel effective pharmacological agents.     

Multiscale and Bayesian approaches to data analysis in genomics high-throughput screening

Tremendous amounts of data are produced by high-throughput screening methods currently employed in drug discovery and product development. A typical cDNA microarray or oligonucleotide-based gene chip experiment easily generates over 10,000 data points for each array or chip. The challenge of inferring meaningful information is formidable given the size and number of these datasets. This paper reviews the current status of statistical tools available for gene expression analysis, with emphasis on Bayesian approaches and multiscale wavelet filtering. Fundamental concepts of Bayesian and multiscale modeling are discussed from the perspective of their potential to address important issues related to the analysis of gene expression data, such as the fact that genomic data often have non-Gaussian distributions and feature localization and multiple scales in both frequency and measurement dimension. Recent publications in these areas are reviewed. Wavelet filtering and the advantages of multiscale methods are demonstrated by application to publicly available gene expression data from the National Cancer Institute (NCI). Multiscale methods, including multiscale principal component analysis (MSPCA), are applied to extract gene subsets and to visualize data in multidimensions for comparisons. Similarity in cell lines and gene selection are effectively visualized and quantitatively compared.