The PowerAtlas: a power and sample size atlas for microarray experimental design and research

Gene Expression Profiles in Children With Suspected Sepsis

STUDY OBJECTIVE

Sepsis recognition is a clinical challenge in children. We aim to determine whether peripheral blood gene expression profiles are associated with pathogen type and sepsis severity in children with suspected sepsis.

METHODS

This was a prospective pilot observational study in a tertiary pediatric emergency department with a convenience sample of children enrolled. Participants were older than 56 days and younger than 18 years, had suspected sepsis, and had not received broad-spectrum antibiotics in the previous 4 hours. Primary outcome was source pathogen, defined as confirmed bacterial source from sterile body fluid or confirmed viral source. Secondary outcome was sepsis severity, defined as maximum therapy required for shock reversal in the first 3 hospital days. We drew peripheral blood for ribonucleic acid isolation at the sepsis protocol activation, obtained gene expression measures with the GeneChip Human Gene 2.0 ST Array, and conducted differential expression analysis.

RESULTS

We collected ribonucleic acid samples from a convenience sample of 122 children with suspected sepsis and 12 healthy controls. We compared the 66 children (54%) with confirmed bacterial or viral infection and found 558 differentially expressed genes, many related to interferon signaling or viral immunity. We did not find statistically significant gene expression differences in patients according to sepsis severity.

CONCLUSION

The study demonstrates feasibility of evaluating gene expression profiling data in children evaluated for sepsis in the pediatric emergency department setting. Our results suggest that gene expression profiling may facilitate identification of source pathogen in children with suspected sepsis, which could ultimately lead to improved tailoring of sepsis treatment and antimicrobial stewardship.

Genome-wide analysis of canine oral malignant melanoma metastasis-associated gene expression

Oral malignant melanoma (OMM) is the most common canine melanocytic neoplasm. Overlap between the somatic mutation profiles of canine OMM and human mucosal melanomas suggest a shared UV-independent molecular aetiology. In common with human mucosal melanomas, most canine OMM metastasise. There is no reliable means of predicting canine OMM metastasis, and systemic therapies for metastatic disease are largely palliative. Herein, we employed exon microarrays for comparative expression profiling of FFPE biopsies of 18 primary canine OMM that metastasised and 10 primary OMM that did not metastasise. Genes displaying metastasis-associated expression may be targets for anti-metastasis treatments, and biomarkers of OMM metastasis. Reduced expression of CXCL12 in the metastasising OMMs implies that the CXCR4/CXCL12 axis may be involved in OMM metastasis. Increased expression of APOBEC3A in the metastasising OMMs may indicate APOBEC3A-induced double-strand DNA breaks and pro-metastatic hypermutation. DNA double strand breakage triggers the DNA damage response network and two Fanconi anaemia DNA repair pathway members showed elevated expression in the metastasising OMMs. Cross-validation was employed to test a Linear Discriminant Analysis classifier based upon the RT-qPCR-measured expression levels of CXCL12, APOBEC3A and RPL29. Classification accuracies of 94% (metastasising OMMs) and 86% (non-metastasising OMMs) were estimated.

Identification of molecular genetic contributants to canine cutaneous mast cell tumour metastasis by global gene expression analysis

Cutaneous mast cell tumours are one of the most common canine cancers. Approximately 25% of the tumours metastasise. Activating c-kit mutations are present in about 20% of tumours, but metastases occur in the absence of mutations. Tumour metastasis is associated with significantly diminished survival in spite of adjuvant chemotherapy. Available prognostic tests do not reliably predict whether a tumour will metastasise. In this study we compared the global expression profiles of 20 primary cutaneous mast cell tumours that metastasised with those of 20 primary tumours that did not metastasise. The objective was to identify genes associated with mast cell tumour metastatic progression that may represent targets for therapeutic intervention and biomarkers for prediction of tumour metastasis. Canine Gene 1.1 ST Arrays were employed for genome-wide expression analysis of formalin-fixed, paraffin-embedded biopsies of mast cell tumours borne by dogs that either died due to confirmed mast cell tumour metastasis, or were still alive more than 1000 days post-surgery. Decreased gene expression in the metastasising tumours appears to be associated with a loss of cell polarity, reduced cell-cell and cell-ECM adhesion, and increased cell deformability and motility. Dysregulated gene expression may also promote extracellular matrix and base membrane degradation, suppression of cell cycle arrest and apoptosis, and angiogenesis. Down-regulation of gene expression in the metastasising tumours may be achieved at least in part by small nucleolar RNA-derived RNA and microRNA-effected gene silencing. Employing cross-validation, a linear discriminant analysis-based classifier featuring 19 genes that displayed two-fold differences in expression between metastasising and non-metastasising tumours was estimated to classify metastasising and non-metastasising tumours with accuracies of 90-100% and 70-100%, respectively. The differential expression of 9 of the discriminator genes was confirmed by quantitative reverse transcription-PCR.

PTSD Blood Transcriptome Mega-Analysis: Shared Inflammatory Pathways across Biological Sex and Modes of Trauma

Transcriptome-wide screens of peripheral blood during the onset and development of posttraumatic stress disorder (PTSD) indicate widespread immune dysregulation. However, little is known as to whether biological sex and the type of traumatic event influence shared or distinct biological pathways in PTSD. We performed a combined analysis of five independent PTSD blood transcriptome studies covering seven types of trauma in 229 PTSD and 311 comparison individuals to synthesize the extant data. Analyses by trauma type revealed a clear pattern of PTSD gene expression signatures distinguishing interpersonal (IP)-related traumas from combat-related traumas. Co-expression network analyses integrated all data and identified distinct gene expression perturbations across sex and modes of trauma in PTSD, including one wound-healing module downregulated in men exposed to combat traumas, one IL-12-mediated signaling module upregulated in men exposed to IP-related traumas, and two modules associated with lipid metabolism and mitogen-activated protein kinase activity upregulated in women exposed to IP-related traumas. Remarkably, a high degree of sharing of transcriptional dysregulation across sex and modes of trauma in PTSD was also observed converging on common signaling cascades, including cytokine, innate immune, and type I interferon pathways. Collectively, these findings provide a broad view of immune dysregulation in PTSD and demonstrate inflammatory pathways of molecular convergence and specificity, which may inform mechanisms and diagnostic biomarkers for the disorder.

Gene expression profiling in colon of mice exposed to food additive titanium dioxide (E171)

Dietary factors that may influence the risks of colorectal cancer, including specific supplements, are under investigation. Previous studies showed the capacity of food additive titanium dioxide (E171) to induce DNA damage in vitro and facilitate growth of colorectal tumours in vivo. This study aimed to investigate the molecular mechanisms behind these effects after E171 exposure. BALB/c mice were exposed by gavage to 5 mg/kg_bw/day of E171 for 2, 7, 14, and 21 days. Transcriptome changes were studied by whole genome mRNA microarray analysis on the mice's distal colons. In addition, histopathological changes as well as a proliferation marker were analysed. The results showed significant gene expression changes in the olfactory/GPCR receptor family, oxidative stress, the immune system and of cancer related genes. Transcriptome analysis also identified genes that thus far have not been included in known biological pathways and can induce functional changes by interacting with other genes involved in different biological pathways. Histopathological analysis showed alteration and disruption in the normal structure of crypts inducing a hyperplastic epithelium. At cell proliferation level, no consistent increase over time was observed. These results may offer a mechanistic framework for the enhanced tumour growth after ingestion of E171 in BALB/c mice.

Unsupervised Placental Gene Expression Profiling Identifies Clinically Relevant Subclasses of Human Preeclampsia

Preeclampsia (PE) is a complex, hypertensive disorder of pregnancy, demonstrating considerable variability in maternal symptoms and fetal outcomes. Unfortunately, prior research has not accounted for this variability, resulting in a lack of robust biomarkers and effective treatments for PE. Here, we created a large (N=330) clinically relevant human placental microarray data set, consisting of 7 previously published studies and 157 highly annotated new samples from a single BioBank. Applying unsupervised clustering to this combined data set identified 3 clinically significant probable etiologies of PE: "maternal", with healthy placentas and term deliveries; "canonical", exhibiting expected clinical, ontological, and histopathologic features of PE; and "immunologic" with severe fetal growth restriction and evidence of maternal antifetal rejection. Moreover, these groups could be distinguished using a small quantitative polymerase chain reaction panel and demonstrated varying influence of maternal factors on PE development. An additional subclass of PE placentas was also revealed to form because of chromosomal abnormalities in these samples, supported by array-based comparative genomic hybridization analysis. Overall, our findings represent a new paradigm in our understanding of the origins and maternal-placental contributions to the pathology of PE. The study of PE represents a unique opportunity to access human tissue associated with a complex hypertensive disorder, and our novel approach could be applied to other hypertensive and heterogeneous human diseases.

Placental transcriptome in development and pathology: expression, function, and methods of analysis

The placenta is the essential organ of mammalian pregnancy and errors in its development and function are associated with a wide range of human pathologies of pregnancy. Genome sequencing has led to methods for investigation of the transcriptome (all expressed RNA species) using microarrays and next-generation sequencing, and implementation of these techniques has identified many novel species of RNA including: micro-RNA, long noncoding RNA, and circular RNA. These species can physically interact with both each other and regulatory proteins to modify gene expression and messenger RNA to protein translation. Transcriptome analysis is actively used to investigate placental development and dysfunction in pathologies ranging from preeclampsia and fetal growth restriction to preterm labor. Genome-wide gene expression analysis is also being applied to identify prognostic and diagnostic biomarkers of these disorders. In this comprehensive review we summarize transcriptome biology, methods of isolation and analysis, application to placental development and pathology, and use in diagnostic analysis in maternal blood. Key information for analysis methods is organized into quick reference tables where current analysis techniques and tools are cited and compared. We have created this review as a practical guide and starting reference for those interested in beginning an investigation into the transcriptome of the placenta.

Fighting a losing battle: vigorous immune response countered by pathogen suppression of host defenses in the chytridiomycosis-susceptible frog Atelopus zeteki

The emergence of the disease chytridiomycosis caused by the chytrid fungus Batrachochytrium dendrobatidis (Bd) has been implicated in dramatic global amphibian declines. Although many species have undergone catastrophic declines and/or extinctions, others appear to be unaffected or persist at reduced frequencies after Bd outbreaks. The reasons behind this variance in disease outcomes are poorly understood: differences in host immune responses have been proposed, yet previous studies suggest a lack of robust immune responses to Bd in susceptible species. Here, we sequenced transcriptomes from clutch-mates of a highly susceptible amphibian, Atelopus zeteki, with different infection histories. We found significant changes in expression of numerous genes involved in innate and inflammatory responses in infected frogs despite high susceptibility to chytridiomycosis. We show evidence of acquired immune responses generated against Bd, including increased expression of immunoglobulins and major histocompatibility complex genes. In addition, fungal-killing genes had significantly greater expression in frogs previously exposed to Bd compared with Bd-naïve frogs, including chitinase and serine-type proteases. However, our results appear to confirm recent in vitro evidence of immune suppression by Bd, demonstrated by decreased expression of lymphocyte genes in the spleen of infected compared with control frogs. We propose susceptibility to chytridiomycosis is not due to lack of Bd-specific immune responses but instead is caused by failure of those responses to be effective. Ineffective immune pathway activation and timing of antibody production are discussed as potential mechanisms. However, in light of our findings, suppression of key immune responses by Bd is likely an important factor in the lethality of this fungus.

Methods for processing microarray data

Quality control must be maintained at every step of a microarray experiment, from RNA isolation through statistical evaluation. Here we provide suggestions for analyzing microarray data. Because the utility of the results depends directly on the design of the experiment, the first critical step is to ensure that the experiment can be properly analyzed and interpreted. What is the biological question? What is the best way to perform the experiment? How many replicates will be required to obtain the desired statistical resolution? Next, the samples must be prepared, pass quality controls for integrity and representation, and be hybridized and scanned. Also, slides with defects, missing data, high background, or weak signal must be rejected. Data from individual slides must be normalized and combined so that the data are as free of systematic bias as possible. The third phase is to apply statistical filters and tests to the data to determine genes (1) expressed above background, (2) whose expression level changes in different samples, and (3) whose RNA-processing patterns or protein associations change. Next, a subset of the data should be validated by an alternative method, such as reverse transcription-polymerase chain reaction (RT-PCR). Provided that this endorses the general conclusions of the array analysis, gene sets whose expression, splicing, polyadenylation, protein binding, etc. change in different samples can be classified with respect to function, sequence motif properties, as well as other categories to extract hypotheses for their biological roles and regulatory logic.

Characterizing differential individual response to porcine reproductive and respiratory syndrome virus infection through statistical and functional analysis of gene expression

We evaluated differences in gene expression in pigs from the Porcine Reproductive and Respiratory Syndrome (PRRS) Host Genetics Consortium initiative showing a range of responses to PRRS virus infection. Pigs were allocated into four phenotypic groups according to their serum viral level and weight gain. RNA obtained from blood at 0, 4, 7, 11, 14, 28, and 42 days post-infection (DPI) was hybridized to the 70-mer 20K Pigoligoarray. We used a blocked reference design for the microarray experiment. This allowed us to account for individual biological variation in gene expression, and to assess baseline effects before infection (0 DPI). Additionally, this design has the flexibility of incorporating future data for differential expression analysis. We focused on evaluating transcripts showing significant interaction of weight gain and serum viral level. We identified 491 significant comparisons [false discovery rate (FDR) = 10%] across all DPI and phenotypic groups. We corroborated the overall trend in direction and level of expression (measured as fold change) at 4 DPI using qPCR (r = 0.91, p ≤ 0.0007). At 4 and 7 DPI, network and functional analyses were performed to assess if immune related gene sets were enriched for genes differentially expressed (DE) across four phenotypic groups. We identified cell death function as being significantly associated (FDR ≤ 5%) with several networks enriched for DE transcripts. We found the genes interferon-alpha 1(IFNA1), major histocompatibility complex, class II, DQ alpha 1 (SLA-DQA1), and major histocompatibility complex, class II, DR alpha (SLA-DRA) to be DE (p ≤ 0.05) between phenotypic groups. Finally, we performed a power analysis to estimate sample size and sampling time-points for future experiments. We concluded the best scenario for investigation of early response to PRRSV infection consists of sampling at 0, 4, and 7 DPI using about 30 pigs per phenotypic group.

Toxicogenomic approaches in developmental toxicology testing

The emergence of toxicogenomic applications provides new tools to characterize, classify, and potentially predict teratogens. However, due to the vast number of experimental and statistical procedural steps, toxicogenomic studies are challenging. Here, we guide researchers through the basic framework of conducting toxicogenomic investigations in the field of developmental toxicology, providing examples of biological and technical factors that may influence response and interpretation. Furthermore, we review current, diverse applications of toxicogenomic-based approaches in teratology testing, including exposure-response characterization (dose and duration), chemical classification studies, and cross-model comparisons study designs. This review is intended to guide scientists through the challenging and complex structure of conducting toxicogenomic analyses, while considering the many applications of using toxicogenomics in study designs and the future of these types of "omics" approaches in developmental toxicology.

Alternative transcription and alternative splicing in cancer

In recent years, the notion of "one gene makes one protein that functions in one signaling pathway" in mammalian cells has been shown to be overly simplistic. Recent genome-wide studies suggest that at least half of the human genes, including many therapeutic target genes, produce multiple protein isoforms through alternative splicing and alternative usage of transcription initiation and/or termination. For example, alternative splicing of the vascular endothelial growth factor gene (VEGFA) produces multiple protein isoforms, which display either pro-angiogenic or anti-angiogenic activities. Similarly, for the majority of human genes, the inclusion or exclusion of exonic sequences enhances the generation of transcript variants and/or protein isoforms that can vary in structure and functional properties. Many of the isoforms produced in this manner are tightly regulated during normal development but are misregulated in cancer cells. Altered expression of transcript variants and protein isoforms for numerous genes is linked with disease and its prognosis, and cancer cells manipulate regulatory mechanisms to express specific isoforms that confer drug resistance and survival advantages. Emerging insights indicate that modulating the expression of transcript and protein isoforms of a gene may hold the key to impeding tumor growth and act as a model for efficient targeting of disease-associated genes at the isoform level. This review highlights the role and regulation of alternative transcription and splicing mechanisms in generating the transcriptome, and the misuse and diagnostic/prognostic potential of alternative transcription and splicing in cancer.

Why is there a lack of consensus on molecular subgroups of glioblastoma? Understanding the nature of biological and statistical variability in glioblastoma expression data

INTRODUCTION

Gene expression patterns characterizing clinically-relevant molecular subgroups of glioblastoma are difficult to reproduce. We suspect a combination of biological and analytic factors confounds interpretation of glioblastoma expression data. We seek to clarify the nature and relative contributions of these factors, to focus additional investigations, and to improve the accuracy and consistency of translational glioblastoma analyses.

METHODS

We analyzed gene expression and clinical data for 340 glioblastomas in The Cancer Genome Atlas (TCGA). We developed a logic model to analyze potential sources of biological, technical, and analytic variability and used standard linear classifiers and linear dimensional reduction algorithms to investigate the nature and relative contributions of each factor.

RESULTS

Commonly-described sources of classification error, including individual sample characteristics, batch effects, and analytic and technical noise make measurable but proportionally minor contributions to inconsistent molecular classification. Our analysis suggests that three, previously underappreciated factors may account for a larger fraction of classification errors: inherent non-linear/non-orthogonal relationships among the genes used in conjunction with classification algorithms that assume linearity; skewed data distributions assumed to be Gaussian; and biologic variability (noise) among tumors, of which we propose three types.

CONCLUSIONS

Our analysis of the TCGA data demonstrates a contributory role for technical factors in molecular classification inconsistencies in glioblastoma but also suggests that biological variability, abnormal data distribution, and non-linear relationships among genes may be responsible for a proportionally larger component of classification error. These findings may have important implications for both glioblastoma research and for translational application of other large-volume biological databases.

Identification of human housekeeping genes and tissue-selective genes by microarray meta-analysis

Background

Categorizing protein-encoding transcriptomes of normal tissues into housekeeping genes and tissue-selective genes is a fundamental step toward studies of genetic functions and genetic associations to tissue-specific diseases. Previous studies have been mainly based on a few data sets with limited samples in each tissue, which restrained the representativeness of their identified genes, and resulted in low consensus among them.

Results

This study compiled 1,431 samples in 43 normal human tissues from 104 microarray data sets. We developed a new method to improve gene expression assessment, and showed that more than ten samples are needed to robustly identify the protein-encoding transcriptome of a tissue. We identified 2,064 housekeeping genes and 2,293 tissue-selective genes, and analyzed gene lists by functional enrichment analysis. The housekeeping genes are mainly involved in fundamental cellular functions, and the tissue-selective genes are strikingly related to functions and diseases corresponding to tissue-origin. We also compared agreements and related functions among our housekeeping genes and those of previous studies, and pointed out some reasons for the low consensuses.

Conclusions

The results indicate that sufficient samples have improved the identification of protein-encoding transcriptome of a tissue. Comprehensive meta-analysis has proved the high quality of our identified HK and TS genes. These results could offer a useful resource for future research on functional and genomic features of HK and TS genes.

Design and validation issues in RNA-seq experiments

The next-generation sequencing technologies are being rapidly applied in biological research. Tens of millions of short sequences generated in a single experiment provide us enormous information on genome composition, genetic variants, gene expression levels and protein binding sites depending on the applications. Various methods are being developed for analyzing the data generated by these technologies. However, the relevant experimental design issues have rarely been discussed. In this review, we use RNA-seq as an example to bring this topic into focus and to discuss experimental design and validation issues pertaining to next-generation sequencing in the quantification of transcripts.

Gene expression profiling identifies inflammation and angiogenesis as distinguishing features of canine hemangiosarcoma

Background

The etiology of hemangiosarcoma remains incompletely understood. Its common occurrence in dogs suggests predisposing factors favor its development in this species. These factors could represent a constellation of heritable characteristics that promote transformation events and/or facilitate the establishment of a microenvironment that is conducive for survival of malignant blood vessel-forming cells. The hypothesis for this study was that characteristic molecular features distinguish hemangiosarcoma from non-malignant endothelial cells, and that such features are informative for the etiology of this disease.

Methods

We first investigated mutations of VHL and Ras family genes that might drive hemangiosarcoma by sequencing tumor DNA and mRNA (cDNA). Protein expression was examined using immunostaining. Next, we evaluated genome-wide gene expression profiling using the Affymetrix Canine 2.0 platform as a global approach to test the hypothesis. Data were evaluated using routine bioinformatics and validation was done using quantitative real time RT-PCR.

Results

Each of 10 tumor and four non-tumor samples analyzed had wild type sequences for these genes. At the genome wide level, hemangiosarcoma cells clustered separately from non-malignant endothelial cells based on a robust signature that included genes involved in inflammation, angiogenesis, adhesion, invasion, metabolism, cell cycle, signaling, and patterning. This signature did not simply reflect a cancer-associated angiogenic phenotype, as it also distinguished hemangiosarcoma from non-endothelial, moderately to highly angiogenic bone marrow-derived tumors (lymphoma, leukemia, osteosarcoma).

Conclusions

The data show that inflammation and angiogenesis are important processes in the pathogenesis of vascular tumors, but a definitive ontogeny of the cells that give rise to these tumors remains to be established. The data do not yet distinguish whether functional or ontogenetic plasticity creates this phenotype, although they suggest that cells which give rise to hemangiosarcoma modulate their microenvironment to promote tumor growth and survival. We propose that the frequent occurrence of canine hemangiosarcoma in defined dog breeds, as well as its similarity to homologous tumors in humans, offers unique models to solve the dilemma of stem cell plasticity and whether angiogenic endothelial cells and hematopoietic cells originate from a single cell or from distinct progenitor cells.

Statistics and bioinformatics in nutritional sciences: analysis of complex data in the era of systems biology

Over the past 2 decades, there have been revolutionary developments in life science technologies characterized by high throughput, high efficiency, and rapid computation. Nutritionists now have the advanced methodologies for the analysis of DNA, RNA, protein, low-molecular-weight metabolites, as well as access to bioinformatics databases. Statistics, which can be defined as the process of making scientific inferences from data that contain variability, has historically played an integral role in advancing nutritional sciences. Currently, in the era of systems biology, statistics has become an increasingly important tool to quantitatively analyze information about biological macromolecules. This article describes general terms used in statistical analysis of large, complex experimental data. These terms include experimental design, power analysis, sample size calculation, and experimental errors (Type I and II errors) for nutritional studies at population, tissue, cellular, and molecular levels. In addition, we highlighted various sources of experimental variations in studies involving microarray gene expression, real-time polymerase chain reaction, proteomics, and other bioinformatics technologies. Moreover, we provided guidelines for nutritionists and other biomedical scientists to plan and conduct studies and to analyze the complex data. Appropriate statistical analyses are expected to make an important contribution to solving major nutrition-associated problems in humans and animals (including obesity, diabetes, cardiovascular disease, cancer, ageing, and intrauterine growth retardation).

Integrative genomic analyses of neurofibromatosis tumours identify SOX9 as a biomarker and survival gene

Understanding the biological pathways critical for common neurofibromatosis type 1 (NF1) peripheral nerve tumours is essential, as there is a lack of tumour biomarkers, prognostic factors and therapeutics. We used gene expression profiling to define transcriptional changes between primary normal Schwann cells (n = 10), NF1-derived primary benign neurofibroma Schwann cells (NFSCs) (n = 22), malignant peripheral nerve sheath tumour (MPNST) cell lines (n = 13), benign neurofibromas (NF) (n = 26) and MPNST (n = 6). Dermal and plexiform NFs were indistinguishable. A prominent theme in the analysis was aberrant differentiation. NFs repressed gene programs normally active in Schwann cell precursors and immature Schwann cells. MPNST signatures strongly differed; genes up-regulated in sarcomas were significantly enriched for genes activated in neural crest cells. We validated the differential expression of 82 genes including the neural crest transcription factor SOX9 and SOX9 predicted targets. SOX9 immunoreactivity was robust in NF and MPSNT tissue sections and targeting SOX9 – strongly expressed in NF1-related tumours – caused MPNST cell death. SOX9 is a biomarker of NF and MPNST, and possibly a therapeutic target in NF1.

Identification of novel candidate genes for treatment response to risperidone and susceptibility for schizophrenia: integrated analysis among pharmacogenomics, mouse expression, and genetic case-control association approaches

BACKGROUND

Pharmacogenomic approaches based on genomewide sets of single nucleotide polymorphisms (SNPs) are now feasible and offer the potential to uncover variants that influence drug response.

METHODS

To detect potential predictor gene variants for risperidone response in schizophrenic subjects, we performed a convergent analysis based on 1) a genomewide (100K SNP) SNP pharmacogenetic study of risperidone response and 2) a global transcriptome study of genes with mRNA levels influenced by risperidone exposure in mouse prefrontal cortex.

RESULTS

Fourteen genes were highlighted as of potential relevance to risperidone activity in both studies: ATP2B2, HS3ST2, UNC5C, BAG3, PDE7B, PAICS, PTGFRN, NR3C2, ZBTB20, ST6GAL2, PIP5K1B, EPHA6, KCNH5, and AJAP1. The SNPs related to these genes that were associated in the pharmacogenetic study were further assessed for evidence for association with schizophrenia in up to three case-control series comprising 1564 cases and 3862 controls in total (Japanese [JPN] 1st and 2nd samples and UK sample). Of 14 SNPs tested, one (rs9389370) in PDE7B showed significant evidence for association with schizophrenia in a discovery sample (p(allele) = .026 in JPN_1st, two-tailed). This finding replicated in a joint analysis of two independent case-control samples (p(JPN_2nd+UK) = .008, one-tailed, uncorrected) and in all combined data sets (p(all) = .0014, two-tailed, uncorrected and p(all) = .018, two-tailed, Bonferroni correction).

CONCLUSIONS

We identified novel candidate genes for treatment response to risperidone and provide evidence that one of these additionally may confer susceptibility to schizophrenia. Specifically, PDE7B is an attractive candidate gene, although evidence from integrated methodology, including pharmacogenomics, pharmacotranscriptomic, and case-control association approaches.

Illustrations on Using the Distribution of a P-value in High Dimensional Data Analyses

Several statistical methods have recently been developed that use the distribution of P-values from multiple tests of hypotheses to analyze data from high-dimensional experiments. These methods are only as valid as the P-values that were derived from test statistics. If an incorrect distribution for a test statistic was used, the P-value will not be valid and the distribution of P-values from multiple test statistics could give misleading results. Moreover, if the correct distribution of a test statistic is used, a distribution of P-values may still give misleading results if P-values are correlated. A primary focus of this paper is on the distribution of a P-value under a null hypothesis, and the test statistic that is considered is the number of rejected null hypotheses. Two issues are demonstrated using six data examples, two that are simulated and four from actual microarray experiments. The results provide some insight into how much of an effect might be introduced into a distribution of P-values if invalid P-values are computed or if P-values are correlated. Additional illustration is given regarding the distribution of a P-value under an alternative hypothesis and some approaches to modeling it are presented.

Comments on the analysis of unbalanced microarray data

MOTIVATION

Permutation testing is very popular for analyzing microarray data to identify differentially expressed (DE) genes; estimating false discovery rates (FDRs) is a very popular way to address the inherent multiple testing problem. However, combining these approaches may be problematic when sample sizes are unequal.

RESULTS

With unbalanced data, permutation tests may not be suitable because they do not test the hypothesis of interest. In addition, permutation tests can be biased. Using biased P-values to estimate the FDR can produce unacceptable bias in those estimates. Results also show that the approach of pooling permutation null distributions across genes can produce invalid P-values, since even non-DE genes can have different permutation null distributions. We encourage researchers to use statistics that have been shown to reliably discriminate DE genes, but caution that associated P-values may be either invalid, or a less-effective metric for discriminating DE genes.

Gene expression profiles of sporadic canine hemangiosarcoma are uniquely associated with breed

The role an individual's genetic background plays on phenotype and biological behavior of sporadic tumors remains incompletely understood. We showed previously that lymphomas from Golden Retrievers harbor defined, recurrent chromosomal aberrations that occur less frequently in lymphomas from other dog breeds, suggesting spontaneous canine tumors provide suitable models to define how heritable traits influence cancer genotypes. Here, we report a complementary approach using gene expression profiling in a naturally occurring endothelial sarcoma of dogs (hemangiosarcoma). Naturally occurring hemangiosarcomas of Golden Retrievers clustered separately from those of non-Golden Retrievers, with contributions from transcription factors, survival factors, and from pro-inflammatory and angiogenic genes, and which were exclusively present in hemangiosarcoma and not in other tumors or normal cells (i.e., they were not due simply to variation in these genes among breeds). Vascular Endothelial Growth Factor Receptor 1 (VEGFR1) was among genes preferentially enriched within known pathways derived from gene set enrichment analysis when characterizing tumors from Golden Retrievers versus other breeds. Heightened VEGFR1 expression in these tumors also was apparent at the protein level and targeted inhibition of VEGFR1 increased proliferation of hemangiosarcoma cells derived from tumors of Golden Retrievers, but not from other breeds. Our results suggest heritable factors mold gene expression phenotypes, and consequently biological behavior in sporadic, naturally occurring tumors.

Transcriptional reprogramming of gene expression in bovine somatic cell chromatin transfer embryos

BACKGROUND

Successful reprogramming of a somatic genome to produce a healthy clone by somatic cells nuclear transfer (SCNT) is a rare event and the mechanisms involved in this process are poorly defined. When serial or successive rounds of cloning are performed, blastocyst and full term development rates decline even further with the increasing rounds of cloning. Identifying the "cumulative errors" could reveal the epigenetic reprogramming blocks in animal cloning.

RESULTS

Bovine clones from up to four generations of successive cloning were produced by chromatin transfer (CT). Using Affymetrix bovine microarrays we determined that the transcriptomes of blastocysts derived from the first and the fourth rounds of cloning (CT1 and CT4 respectively) have undergone an extensive reprogramming and were more similar to blastocysts derived from in vitro fertilization (IVF) than to the donor cells used for the first and the fourth rounds of chromatin transfer (DC1 and DC4 respectively). However a set of transcripts in the cloned embryos showed a misregulated pattern when compared to IVF embryos. Among the genes consistently upregulated in both CT groups compared to the IVF embryos were genes involved in regulation of cytoskeleton and cell shape. Among the genes consistently upregulated in IVF embryos compared to both CT groups were genes involved in chromatin remodelling and stress coping.

CONCLUSION

The present study provides a data set that could contribute in our understanding of epigenetic errors in somatic cell chromatin transfer. Identifying "cumulative errors" after serial cloning could reveal some of the epigenetic reprogramming blocks shedding light on the reprogramming process, important for both basic and applied research.

Challenges and approaches to statistical design and inference in high-dimensional investigations

Advances in modern technologies have facilitated high-dimensional experiments (HDEs) that generate tremendous amounts of genomic, proteomic, and other "omic" data. HDEs involving whole-genome sequences and polymorphisms, expression levels of genes, protein abundance measurements, and combinations thereof have become a vanguard for new analytic approaches to the analysis of HDE data. Such situations demand creative approaches to the processes of statistical inference, estimation, prediction, classification, and study design. The novel and challenging biological questions asked from HDE data have resulted in many specialized analytic techniques being developed. This chapter discusses some of the unique statistical challenges facing investigators studying high-dimensional biology and describes some approaches being developed by statistical scientists. We have included some focus on the increasing interest in questions involving testing multiple propositions simultaneously, appropriate inferential indicators for the types of questions biologists are interested in, and the need for replication of results across independent studies, investigators, and settings. A key consideration inherent throughout is the challenge in providing methods that a statistician judges to be sound and a biologist finds informative.

Oligonucleotide microarrays for bacteriophage expression studies

Gene expression microarrays offer the ability to monitor the expression of all phage genes over an infection cycle. However, there are relatively few reports to date of microarrays being used to investigate phage biology. This chapter aims to provide an overview of how to design and implement a microarray experiment to investigate phage biology. Given the nature of microarrays being specific to an organism, each will provide a number of unique issues. In this chapter, we outline the basic theory behind microarrays and provide details on how to implement a microarray experiment from the design of oligonucleotide probes through to the hybridisation of microarrays. The matter of designing oligonucleotide probes will be discussed with regards to how probe length, secondary structure, free energy, probe orientation and amplification all have to be taken into account. As means of an example, the conditions used for the hybridisation of an array designed to be specific to the cyanophage S-PM2 is detailed.

Effects of spaceflight on murine skeletal muscle gene expression

Spaceflight results in a number of adaptations to skeletal muscle, including atrophy and shifts toward faster muscle fiber types. To identify changes in gene expression that may underlie these adaptations, we used both microarray expression analysis and real-time polymerase chain reaction to quantify shifts in mRNA levels in the gastrocnemius from mice flown on the 11-day, 19-h STS-108 shuttle flight and from normal gravity controls. Spaceflight data also were compared with the ground-based unloading model of hindlimb suspension, with one group of pure suspension and one of suspension followed by 3.5 h of reloading to mimic the time between landing and euthanization of the spaceflight mice. Analysis of microarray data revealed that 272 mRNAs were significantly altered by spaceflight, the majority of which displayed similar responses to hindlimb suspension, whereas reloading tended to counteract these responses. Several mRNAs altered by spaceflight were associated with muscle growth, including the phosphatidylinositol 3-kinase regulatory subunit p85alpha, insulin response substrate-1, the forkhead box O1 transcription factor, and MAFbx/atrogin1. Moreover, myostatin mRNA expression tended to increase, whereas mRNA levels of the myostatin inhibitor FSTL3 tended to decrease, in response to spaceflight. In addition, mRNA levels of the slow oxidative fiber-associated transcriptional coactivator peroxisome proliferator-associated receptor (PPAR)-gamma coactivator-1alpha and the transcription factor PPAR-alpha were significantly decreased in spaceflight gastrocnemius. Finally, spaceflight resulted in a significant decrease in levels of the microRNA miR-206. Together these data demonstrate that spaceflight induces significant changes in mRNA expression of genes associated with muscle growth and fiber type.

A comparison of molecular alterations in environmental and genetic rat models of ADHD: a pilot study

Attention deficit hyperactivity disorder (ADHD) is the most common neurobehavioral disorder in school-aged children. In addition to genetic factors, environmental influences or gene x environmental interactions also play an important role in ADHD. One example of a well studied environmental risk factor for ADHD is exposure to polychlorinated biphenyls (PCBs). In this study, we investigated whether the well-established genetic model of ADHD based on the spontaneously hypertensive rat (SHR) and a well established PCB-based model of ADHD exhibited similar molecular changes in brain circuits involved in ADHD. The brains from 28 male rats (8 SHR, 8 Sprague-Dawley (SD) controls, 8 Wistar/Kyoto (WKY) controls, and 4 PCB-exposed SD rats) were harvested at postnatal days (PNDs) 55-65 and RNA was isolated from six brain regions of interest. The RNA was analyzed for differences in expression of a set of 308 probe sets interrogating 218 unique genes considered highly relevant to ADHD or epigenetic gene regulation using the Rat RAE230 2.0 GeneChip (Affymetrix). Selected observations were confirmed by real-time quantitative RT-PCR. The results show that the expression levels of genes Gnal, COMT, Adrbk1, Ntrk2, Hk1, Syt11, and Csnk1a1 were altered in both the SHR rats and the PCB-exposed SD rats. Arrb2, Stx12, Aqp6, Syt1, Ddc, and Pgk1 expression levels were changed only in the PCB-exposed SD rats. Genes with altered expression only in the SHRs included Oprm1, Calcyon, Calmodulin, Lhx1, and Hes6. The epigenetic genes Crebbp, Mecp2, and Hdac5 are significantly altered in both models. The data provide strong evidence that genes and environment can affect different set of genes in two different models of ADHD and yet result in the similar disease-like symptoms.

Power enhancement via multivariate outlier testing with gene expression arrays

Motivation: As the use of microarrays in human studies continues to increase, stringent quality assurance is necessary to ensure accurate experimental interpretation. We present a formal approach for microarray quality assessment that is based on dimension reduction of established measures of signal and noise components of expression followed by parametric multivariate outlier testing.

Results: We applied our approach to several data resources. First, as a negative control, we found that the Affymetrix and Illumina contributions to MAQC data were free from outliers at a nominal outlier flagging rate of α=0.01. Second, we created a tunable framework for artificially corrupting intensity data from the Affymetrix Latin Square spike-in experiment to allow investigation of sensitivity and specificity of quality assurance (QA) criteria. Third, we applied the procedure to 507 Affymetrix microarray GeneChips processed with RNA from human peripheral blood samples. We show that exclusion of arrays by this approach substantially increases inferential power, or the ability to detect differential expression, in large clinical studies.

Availability:http://bioconductor.org/packages/2.3/bioc/html/arrayMvout.html and http://bioconductor.org/packages/2.3/bioc/html/affyContam.html affyContam (credentials: readonly/readonly)

Contact:aasare@immunetolerance.org; stvjc@channing.harvard.edu

Evaluating statistical methods using plasmode data sets in the age of massive public databases: an illustration using false discovery rates

Plasmode is a term coined several years ago to describe data sets that are derived from real data but for which some truth is known. Omic techniques, most especially microarray and genomewide association studies, have catalyzed a new zeitgeist of data sharing that is making data and data sets publicly available on an unprecedented scale. Coupling such data resources with a science of plasmode use would allow statistical methodologists to vet proposed techniques empirically (as opposed to only theoretically) and with data that are by definition realistic and representative. We illustrate the technique of empirical statistics by consideration of a common task when analyzing high dimensional data: the simultaneous testing of hundreds or thousands of hypotheses to determine which, if any, show statistical significance warranting follow-on research. The now-common practice of multiple testing in high dimensional experiment (HDE) settings has generated new methods for detecting statistically significant results. Although such methods have heretofore been subject to comparative performance analysis using simulated data, simulating data that realistically reflect data from an actual HDE remains a challenge. We describe a simulation procedure using actual data from an HDE where some truth regarding parameters of interest is known. We use the procedure to compare estimates for the proportion of true null hypotheses, the false discovery rate (FDR), and a local version of FDR obtained from 15 different statistical methods.

Plasmode is a term used to describe a data set that has been derived from real data but for which some truth is known. Statistical methods that analyze data from high dimensional experiments (HDEs) seek to estimate quantities that are of interest to scientists, such as mean differences in gene expression levels and false discovery rates. The ability of statistical methods to accurately estimate these quantities depends on theoretical derivations or computer simulations. In computer simulations, data for which the true value of a quantity is known are often simulated from statistical models, and the ability of a statistical method to estimate this quantity is evaluated on the simulated data. However, in HDEs there are many possible statistical models to use, and which models appropriately produce data that reflect properties of real data is an open question. We propose the use of plasmodes as one answer to this question. If done carefully, plasmodes can produce data that reflect reality while maintaining the benefits of simulated data. We show one method of generating plasmodes and illustrate their use by comparing the performance of 15 statistical methods for estimating the false discovery rate in data from an HDE.

Microarray gene-expression profiling to predict lymph node metastasis in penile carcinoma

OBJECTIVE

To determine the value of gene-expression profiling as a predictor of the status of the regional nodes in patients with penile carcinoma.

PATIENTS AND METHODS

Tumour samples of 56 patients with penile squamous cell carcinoma were analysed for the gene expression on 35 k oligoarrays; 32 were from patients with histopathologically confirmed lymph node metastases and 24 from those with no lymph node involvement. The 56 patients were divided into a training and validation set. For the training set, 15 patients with histopathologically confirmed nodal metastases and 15 without were selected. The validation set consisted of the remaining 26 patients, containing 17 node-positive and nine with no nodal metastases.

RESULTS

A 44-probe classifier had the best performance within the training set; this classifier correctly assigned 29 of 30 specimens in the training set to the two outcome groups. In the validation set of 26 tumours, the classifier correctly assigned 14 of the 26 (54%) specimens to the two outcome groups. Of the 17 specimens with histologically confirmed nodal involvement, 12 were classified as node-positive and five as node-negative, resulting in a sensitivity of 71%. Of the nine specimens from node-negative patients, two were correctly classified as node-negative and seven as node positive, resulting in a specificity of 22%.

CONCLUSIONS

In this series, gene expression profiling did not produce a useful classifier to predict nodal involvement in patients with penile carcinoma.

Machine learning-based receiver operating characteristic (ROC) curves for crisp and fuzzy classification of DNA microarrays in cancer research

Receiver operating characteristic (ROC) curves were generated to obtain classification area under the curve (AUC) as a function of feature standardization, fuzzification, and sample size from nine large sets of cancer-related DNA microarrays. Classifiers used included k nearest neighbor (kNN), näive Bayes classifier (NBC), linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), learning vector quantization (LVQ1), logistic regression (LOG), polytomous logistic regression (PLOG), artificial neural networks (ANN), particle swarm optimization (PSO), constricted particle swarm optimization (CPSO), kernel regression (RBF), radial basis function networks (RBFN), gradient descent support vector machines (SVMGD), and least squares support vector machines (SVMLS). For each data set, AUC was determined for a number of combinations of sample size, total sum[-log(p)] of feature t-tests, with and without feature standardization and with (fuzzy) and without (crisp) fuzzification of features. Altogether, a total of 2,123,530 classification runs were made. At the greatest level of sample size, ANN resulted in a fitted AUC of 90%, while PSO resulted in the lowest fitted AUC of 72.1%. AUC values derived from 4NN were the most dependent on sample size, while PSO was the least. ANN depended the most on total statistical significance of features used based on sum[-log(p)], whereas PSO was the least dependent. Standardization of features increased AUC by 8.1% for PSO and -0.2% for QDA, while fuzzification increased AUC by 9.4% for PSO and reduced AUC by 3.8% for QDA. AUC determination in planned microarray experiments without standardization and fuzzification of features will benefit the most if CPSO is used for lower levels of feature significance (i.e., sum[-log(p)] ~ 50) and ANN is used for greater levels of significance (i.e., sum[-log(p)] ~ 500). When only standardization of features is performed, studies are likely to benefit most by using CPSO for low levels of feature statistical significance and LVQ1 for greater levels of significance. Studies involving only fuzzification of features should employ LVQ1 because of the substantial gain in AUC observed and low expense of LVQ1. Lastly, PSO resulted in significantly greater levels of AUC (89.5% average) when feature standardization and fuzzification were performed. In consideration of the data sets used and factors influencing AUC which were investigated, if low-expense computation is desired then LVQ1 is recommended. However, if computational expense is of less concern, then PSO or CPSO is recommended.

Accommodating pathway information in expression quantitative trait locus analysis

The availability of high-throughput genotyping technologies and microarray assays has allowed researchers to consider pursuing investigations whose ultimate goal is the identification of genetic variations that influence levels of gene expression, e.g., "expression quantitative trait locus" or "eQTL" mapping studies. However, the large number of genes whose expression levels can be tested for association with genetic variations in such studies can create both statistical and biological interpretive problems. We consider the integrated analysis of eQTL mapping data that incorporates pathway, function, and disease process information. The goal of this analysis is to determine if compelling patterns emerge from the data that are consistent with the notion that perturbations in the molecular physiologic environment induced by genetic variations implicate the expression patterns of multiple genes via genetic network relationships or feedback mechanisms. We apply available genetic network and pathway analysis software, as well as a novel regression analysis technique, to carry out the proposed studies. We also consider extensions of the proposed strategies and areas of future research.

Microarray analysis

Microarrays and related technologies have allowed investigators to ask biological questions in far greater detail than has previously been possible. Microarrays had a troubled beginning, but most of these problems resulted from the growing pains of this technology, which, like many new things, was initially more promise than delivery. Nevertheless, over the past few years, investigators have learned how to achieve optimal performance of technology, and now exciting discoveries are made using microarray-based research. Many of the advances have come from the realization that microarrays are not a magic tool but rather are like any other measurement device. Unless microarray experimentation is coupled with good experimental practices, it will not yield valid results or, worse yet, may lead to misleading results. In this chapter, we highlight some of the important steps that should be taken to successfully conduct a microarray study. These steps include a clearly stated biological question, experimental design, careful experimental conduct, complete statistical analysis, validation/verification of results, and dissemination of the data.

Wide gene expression profiling of ischemia-reperfusion injury in human liver transplantation

Ischemia-reperfusion injury (IRI) causes up to 10% of early liver failures in humans and can lead to a higher incidence of acute and chronic rejection. So far, very few studies have investigated wide gene expression profiles associated with the IRI process. The discovery of novel genes activated by IRI might lead to the identification of potential target genes for the prevention or treatment of the injury. In our study, we compared gene expression levels in reperfused livers (RL group) vs. the basal values before retrieval from the donor (basal liver [BL] group) using oligonucleotide array technology. We examined 10 biopsies from 5 livers, analyzing approximately 33,000 genes represented on the Affymetrix HG-U133APlus 2.0 oligonucleotide arrays (Affymetrix, Santa Clara, CA). About 13,000 individual genes were considered expressed in at least 1 condition. A total of 795 genes whose expression is significantly modified by ischemia-reperfusion in human liver transplantation were identified in this study. Some of them are likely to be completely activated by IRI, as they are not expressed in basal livers. The supervised gene expression analysis revealed that at least 12% of the genes involved in the apoptotic process, 12.5% of the genes involved in inflammatory processes, and 22.5% of the genes encoding for heat shock proteins are differentially expressed in RL samples vs. BL samples. Furthermore, IRI induces the upregulation of some genes' coding for adhesion molecules and integrins. In conclusion, we have identified a relevant amount of early genes regulated in the human liver after 7-9 hours of cold ischemia and 2 hours from reperfusion, many of them not having been described before in this process. Their analyses may help us to better understand the pathophysiology of IRI and to characterize potential target genes for the prevention or treatment of the liver injury in order to increase the number of patients that successfully undergo transplantation.

Using RNA sample titrations to assess microarray platform performance and normalization techniques

We have assessed the utility of RNA titration samples for evaluating microarray platform performance and the impact of different normalization methods on the results obtained. As part of the MicroArray Quality Control project, we investigated the performance of five commercial microarray platforms using two independent RNA samples and two titration mixtures of these samples. Focusing on 12,091 genes common across all platforms, we determined the ability of each platform to detect the correct titration response across the samples. Global deviations from the response predicted by the titration ratios were observed. These differences could be explained by variations in relative amounts of messenger RNA as a fraction of total RNA between the two independent samples. Overall, both the qualitative and quantitative correspondence across platforms was high. In summary, titration samples may be regarded as a valuable tool, not only for assessing microarray platform performance and different analysis methods, but also for determining some underlying biological features of the samples.

Microarray data on gene modulation by HIV-1 in immune cells: 2000-2006

Here, we review 34 HIV microarray studies in human immune cells over the period of 2000-March 2006 with emphasis on analytical approaches used and conceptual advances on HIV modulation of target cells (CD4 T cell, macrophage) and nontargets such as NK cell, B cell, and dendritic cell subsets. Results to date address advances on gene modulation associated with immune dysregulation, susceptibility to apoptosis, virus replication, and viral persistence following in vitro or in vivo infection/exposure to HIV-1 virus or HIV-1 accessory proteins. In addition to gene modulation associated with known functional correlates of HIV infection and replication (e.g., T cell apoptosis), microarray data have yielded novel, potential mechanisms of HIV-mediated pathogenesis such as modulation of cholesterol biosynthetic genes in CD4 T cells (relevant to virus replication and infectivity) and modulation of proteasomes and histone deacetylases in chronically infected cell lines (relevant to virus latency). Intrinsic challenges in summarizing gene modulation studies remain in development of sound approaches for comparing data obtained using different platforms and analytical tools, deriving unifying concepts to distil the large volumes of data collected, and the necessity to impose a focus for validation on a small fraction of genes. Notwithstanding these challenges, the field overall continues to demonstrate progress in expanding the pool of target genes validated to date in in vitro and in vivo datasets and understanding the functional correlates of gene modulation to HIV-1 pathogenesis in vivo.