Person:

Carey, Vincent

Background: Graphical models (e.g., Bayesian networks) have been used frequently to describe complex interaction patterns and dependent structures among genes and other phenotypes. Estimation of such networks has been a challenging problem when the genes considered greatly outnumber the samples, and the situation is exacerbated when one wishes to consider the impact of polymorphisms (SNPs) in genes. Results: Here we describe a multistep approach to infer a gene-SNP network from gene expression and genotyped SNP data. Our approach is based on 1) construction of a graphical Gaussian model (GGM) based on small sample estimation of partial correlation and false-discovery rate multiple testing; 2) extraction of a subnetwork of genes directly linked to a target candidate gene of interest; 3) identification of cis-acting regulatory variants for the genes composing the subnetwork; and 4) evaluating the identified cis-acting variants for trans-acting regulatory effects of the target candidate gene. This approach identifies significant gene-gene and gene-SNP associations not solely on the basis of gene co-expression but rather through whole-network modeling. We demonstrate the method by building two complex gene-SNP networks around Interferon Receptor 12B2 (IL12RB2) and Interleukin 1B (IL1B), two biologic candidates in asthma pathogenesis, using 534,290 genotyped variants and gene expression data on 22,177 genes from total RNA derived from peripheral blood CD4+ lymphocytes from 154 asthmatics. Conclusion: Our results suggest that graphical models based on integrative genomic data are computationally efficient, work well with small samples, and can describe complex interactions among genes and polymorphisms that could not be identified by pair-wise association testing.

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.

Background: With the recent development of microarray technologies, the comparability of gene expression data obtained from different platforms poses an important problem. We evaluated two widely used platforms, Affymetrix U133 Plus 2.0 and the Illumina HumanRef-8 v2 Expression Bead Chips, for comparability in a biological system in which changes may be subtle, namely fetal lung tissue as a function of gestational age. Results: We performed the comparison via sequence-based probe matching between the two platforms. "Significance grouping" was defined as a measure of comparability. Using both expression correlation and significance grouping as measures of comparability, we demonstrated that despite overall cross-platform differences at the single gene level, increased correlation between the two platforms was found in genes with higher expression level, higher probe overlap, and lower p-value. We also demonstrated that biological function as determined via KEGG pathways or GO categories is more consistent across platforms than single gene analysis. Conclusion: We conclude that while the comparability of the platforms at the single gene level may be increased by increasing sample size, they are highly comparable ontologically even for subtle differences in a relatively small sample size. Biologically relevant inference should therefore be reproducible across laboratories using different platforms.

Summary: The rtracklayer package supports the integration of existing genome browsers with experimental data analyses performed in R. The user may (i) transfer annotation tracks to and from a genome browser and (ii) create and manipulate browser views to focus on a particular set of annotations in a specific genomic region. Currently, the UCSC genome browser is supported. Availability: The package is freely available from http://www.bioconductor.org/. A quick-start vignette is included with the package. Contact: mflawren@fhcrc.org

Background: Comparative characterization of genome-wide transcriptional changes during infection can help elucidate the mechanisms underlying host susceptibility. In this study, transcriptional profiling of the mouse colon was carried out in two cognate lines of mice that differ in their response to Citrobacter rodentium infection; susceptible inbred FVB/N and resistant outbred Swiss Webster mice. Gene expression in the distal colon was determined prior to infection, and at four and nine days post-inoculation using a whole mouse genome Affymetrix array. Results: Computational analysis identified 462 probe sets more than 2-fold differentially expressed between uninoculated resistant and susceptible mice. In response to C. rodentium infection, 5,123 probe sets were differentially expressed in one or both lines of mice. Microarray data were validated by quantitative real-time RT-PCR for 35 selected genes and were found to have a 94% concordance rate. Transcripts represented by 1,547 probe sets were differentially expressed between susceptible and resistant mice regardless of infection status, a host effect. Genes associated with transport were over-represented to a greater extent than even immune responserelated genes. Electrolyte analysis revealed reduction in serum levels of chloride and sodium in susceptible animals. Conclusion: The results support the hypothesis that mortality in C. rodentium-infected susceptible mice is associated with impaired intestinal ion transport and development of fatal fluid loss and dehydration. These studies contribute to our understanding of the pathogenesis of C. rodentium and suggest novel strategies for the prevention and treatment of diarrhea associated with intestinal bacterial infections.

Bronchodilator response (BDR) is an important asthma phenotype that measures reversibility of airway obstruction by comparing lung function (i.e. FEV1) before and after the administration of a short-acting β2-agonist, the most common rescue medications used for the treatment of asthma. BDR also serves as a test of β2-agonist efficacy. BDR is a complex trait that is partly under genetic control. A genome-wide association study (GWAS) of BDR, quantified as percent change in baseline FEV1 after administration of a β2-agonist, was performed with 1,644 non-Hispanic white asthmatic subjects from six drug clinical trials: CAMP, LOCCS, LODO, a medication trial conducted by Sepracor, CARE, and ACRN. Data for 469,884 single-nucleotide polymorphisms (SNPs) were used to measure the association of SNPs with BDR using a linear regression model, while adjusting for age, sex, and height. Replication of primary P-values was attempted in 501 white subjects from SARP and 550 white subjects from DAG. Experimental evidence supporting the top gene was obtained via siRNA knockdown and Western blotting analyses. The lowest overall combined P-value was 9.7E-07 for SNP rs295137, near the SPATS2L gene. Among subjects in the primary analysis, those with rs295137 TT genotype had a median BDR of 16.0 (IQR = [6.2, 32.4]), while those with CC or TC genotypes had a median BDR of 10.9 (IQR = [5.0, 22.2]). SPATS2L mRNA knockdown resulted in increased β2-adrenergic receptor levels. Our results suggest that SPATS2L may be an important regulator of β2-adrenergic receptor down-regulation and that there is promise in gaining a better understanding of the biological mechanisms of differential response to β2-agonists through GWAS.

Recent breakthroughs in next-generation sequencing technologies allow cost-effective methods for measuring a growing list of cellular properties, including DNA sequence and structural variation. Next-generation sequencing has the potential to revolutionize complex trait genetics by directly measuring common and rare genetic variants within a genome-wide context. Because for a given gene both rare and common causal variants can coexist and have independent effects on a trait, strategies that model the effects of both common and rare variants could enhance the power of identifying disease-associated genes. To date, little work has been done on integrating signals from common and rare variants into powerful statistics for finding disease genes in genome-wide association studies. In this analysis of the Genetic Analysis Workshop 17 data, we evaluate various strategies for association of rare, common, or a combination of both rare and common variants on quantitative phenotypes in unrelated individuals. We show that the analysis of common variants only using classical approaches can achieve higher power to detect causal genes than recently proposed rare variant methods and that strategies that combine association signals derived independently in rare and common variants can slightly increase the power compared to strategies that focus on the effect of either the rare variants or the common variants.

Background: Smoking while pregnant is associated with a myriad of negative health outcomes in the child. Some of the detrimental effects may be due to epigenetic modifications, although few studies have investigated this hypothesis in detail. Objectives: To characterize site-specific epigenetic modifications conferred by prenatal smoking exposure within asthmatic children. Methods: Using Illumina HumanMethylation27 microarrays, we estimated the degree of methylation at 27,578 distinct DNA sequences located primarily in gene promoters using whole blood DNA samples from the Childhood Asthma Management Program (CAMP) subset of Asthma BRIDGE childhood asthmatics (n = 527) ages 5–12 with prenatal smoking exposure data available. Using beta-regression, we screened loci for differential methylation related to prenatal smoke exposure, adjusting for gender, age and clinical site, and accounting for multiple comparisons by FDR. Results: Of 27,578 loci evaluated, 22,131 (80%) passed quality control assessment and were analyzed. Sixty-five children (12%) had a history of prenatal smoke exposure. At an FDR of 0.05, we identified 19 CpG loci significantly associated with prenatal smoke, of which two replicated in two independent populations. Exposure was associated with a 2% increase in mean CpG methylation in FRMD4A (p = 0.01) and Cllorf52 (p = 0.001) compared to no exposure. Four additional genes, XPNPEP1, PPEF2, SMPD3 and CRYGN, were nominally associated in at least one replication group. Conclusions: These data suggest that prenatal exposure to tobacco smoke is associated with reproducible epigenetic changes that persist well into childhood. However, the biological significance of these altered loci remains unknown.

We investigate 71 single nucleotide polymorphisms (SNPs) identified in meta-analytic studies of genome-wide association studies (GWAS) of breast cancer, the majority of which are located in intergenic or intronic regions. To explore regulatory impacts of these variants we conducted expression quantitative loci (eQTL) analyses on tissue samples from 376 invasive postmenopausal breast cancer cases in the Nurses’ Health Study (NHS) diagnosed from 1990–2004. Expression analysis was conducted on all formalin-fixed paraffin-embedded (FFPE) tissue samples (and on 264 adjacent normal samples) using the Affymetrix Human Transcriptome Array. Significance and ranking of associations between tumor receptor status and expression variation was preserved between NHS FFPE and TCGA fresh-frozen sample sets (Spearman r = 0.85, p<10^-10 for 17 of the 21 Oncotype DX recurrence signature genes). At an FDR threshold of 10%, we identified 27 trans-eQTLs associated with expression variation in 217 distinct genes. SNP-gene associations can be explored using an open-source interactive browser distributed in a Bioconductor package. Using a new a procedure for testing hypotheses relating SNP content to expression patterns in gene sets, defined as molecular function pathways, we find that loci on 6q14 and 6q25 affect various gene sets and molecular pathways (FDR < 10%). Although the ultimate biological interpretation of the GWAS-identified variants remains to be uncovered, this study validates the utility of expression analysis of this FFPE expression set for more detailed integrative analyses.

Human-associated microbial communities have a crucial role in determining our health and well-being and this has led to the continuing development of microbiome-based therapies such as faecal microbiota transplantation. These microbial communities are very complex, dynamic and highly personalized ecosystems exhibiting a high degree of inter-individual variability in both species assemblages and abundance profiles9. It is not known whether the underlying ecological dynamics of these communities, which can be parameterized by growth rates, and intra- and inter-species interactions in population dynamics models, are largely host-independent (that is, universal) or host-specific. If the inter-individual variability reflects host-specific dynamics due to differences in host lifestyle, physiology or genetics, then generic microbiome manipulations may have unintended consequences, rendering them ineffective or even detrimental. Alternatively, microbial ecosystems of different subjects may exhibit universal dynamics, with the inter-individual variability mainly originating from differences in the sets of colonizing species Here we develop a new computational method to characterize human microbial dynamics. By applying this method to cross-sectional data from two large-scale metagenomic studies—the Human Microbiome Project and the Student Microbiome Project—we show that gut and mouth microbiomes display pronounced universal dynamics, whereas communities associated with certain skin sites are probably shaped by differences in the host environment. Notably, the universality of gut microbial dynamics is not observed in subjects with recurrent Clostridium difficile infection but is observed in the same set of subjects after faecal microbiota transplantation. These results fundamentally improve our understanding of the processes that shape human microbial ecosystems, and pave the way to designing general microbiome-based therapies.