Multiple Testing Correction Methods Research Articles

Causality of the gut microbiome and atherosclerosis-related lipids: a bidirectional Mendelian Randomization study

AimsRecent studies have indicated an association between intestinal flora and lipids. However, observational studies cannot indicate causality. In this study, we aimed to investigate the potentially causal relationships between the intestinal flora and blood lipids.MethodsWe performed a bidirectional two-sample Mendelian Randomization (MR) analysis to investigate the causal relationship between intestinal flora and blood lipids. Summary statistics of genome-wide association studies (GWASs) for the 211 intestinal flora and blood lipid traits (n = 5) were obtained from public datasets. Five recognized MR methods were applied to assess the causal relationship with lipids, among which, the inverse-variance weighted (IVW) regression was used as the primary MR method. A series of sensitivity analyses were performed to test the robustness of the causal estimates.ResultsThe results indicated a potential causal association between 19 intestinal flora and dyslipidemia in humans. Genus Ruminococcaceae, Christensenellaceae, Parasutterella, Terrisporobacter, Parabacteroides, Class Erysipelotrichia, Family Erysipelotrichaceae, and order Erysipelotrichales were associated with higher dyslipidemia, whereas genus Oscillospira, Peptococcus, Ruminococcaceae UCG010, Ruminococcaceae UCG011, Dorea, and Family Desulfovibrionaceae were associated with lower dyslipidemia. After using the Bonferroni method for multiple testing correction, Only Desulfovibrionaceae [Estimate = -0.0418, 95% confidence interval [CI]: 0.9362–0.9826, P = 0.0007] exhibited stable and significant negative associations with ApoB levels. The inverse MR analysis did not find a significant causal effect of lipids on the intestinal flora. Additionally, no significant heterogeneity or horizontal pleiotropy for IVs was observed in the analysis.ConclusionThe study suggested a causal relationship between intestinal flora and dyslipidemia. These findings will provide a meaningful reference to discover dyslipidemia for intervention to address the problems in the clinic.

The associations between gut microbiota and inflammatory skin diseases: a bi-directional two-sample Mendelian randomization study.

Accumulating evidence shows that dysregulation of intestinal flora is associated with inflammatory skin diseases, specifically atopic dermatitis (AD), psoriasis (PSO), and rosacea (ROS). However, the causality is still unclear. To study the underlying causality between gut microbiota (GM) and AD, PSO, and ROS, a bi-directional two-sample Mendelian randomization (2SMR) analysis was conducted. Summary statistics of gut microbiota, AD, PSO, and ROS were extracted from large-scale genome-wide association studies (GWASs). In 2SMR analysis, in addition to the inverse variance weighted as the principal method for evaluating causal association, four different methods were also used. Sensitivity analysis and reverse 2SMR study were implemented to evaluate the robustness of 2SMR results or reverse causal relationship, respectively. A total of 24 specific gut microbiota species related to AD, PSO, and ROS were identified by 2SMR analysis. After using the Bonferroni method for multiple testing correction, family FamilyXIII (ID: 1957) [OR = 1.28 (1.13, 1.45), p = 9.26e-05] and genus Eubacteriumfissicatenagroup (ID: 14373) [OR = 1.20 (1.09, 1.33), p = 1.65e-04] were associated with an increased risk for AD and PSO, respectively. The genus Dialister showed a negative association, suggesting a protective role against both atopic dermatitis and rosacea. Our reverse 2SMR analysis indicated no reverse causality between these inflammatory skin diseases and the identified gut microbiota. In summary, this study provided evidence for the causality between GM and inflammatory skin diseases. These findings suggested that supplementing specific bacterial taxa may be an effective therapy for AD, PSO, and ROS.

Relationship between gut microbiota and thyroid function: a two-sample Mendelian randomization study.

Numerous observational studies have indicated a link between the composition of gut microbiota and thyroid function. Nevertheless, the precise causal relationship between gut microbiota and thyroid function remains uncertain. In this two-sample Mendelian randomization study, we utilized summary data from a genome-wide association study of gut microbiota composition in 18,340 participants from 24 cohorts, as well as summary statistics on thyroid hormones and thyroid-stimulating hormone from the ThyroidOmics Consortium and summary statistics on hypothyroidism and hyperthyroidism from the FinnGen R8 release. Five different methods, including inverse variance weighting, MR-Egger, weighted median, weighted mode, and simple mode, were employed to examine the causal relationship between gut microbiota and thyroid function. Reverse Mendelian randomization analysis was conducted for taxa identified as having a causal relationship with thyroid function in the Mendelian randomization analysis. To assess the robustness of the results, sensitivity analyses were conducted employing Cochran's Q test, MR-Egger intercept test, MR-PRESSO global test, and leave-one-out analysis. Through MR analysis of 211 microbial taxa and 4 phenotypes, we identified a total of 34 gut microbiota taxa that were associated with the outcomes. After using the bonferroni method for multiple testing correction, phylum Actinobacteria (id.400) had a protective effect on hypothyroidism (OR=0.883, 95% CI: 0.817-0.955, P=0.002), and class Deltaproteobacteria (id.3087) had a protective effect on hyperthyroidism (OR=0.549, 95% CI: 0.374-0.805, P=0.002). According to the results of reverse MR analysis, no significant causal effect of the four phenotypes was found on gut microbiota. No significant horizontal pleiotropy was detected based on MR-Egger intercept test and MR-PRESSO global test. Through two-sample MR analysis, we identified specific gut microbiota taxa at the genetic level that are predicted to have a causal relationship with thyroid function, which may serve as useful biomarkers for early disease diagnosis.

Roles of gut microbiome in epilepsy risk: A Mendelian randomization study.

Recent studies have suggested an association between gut microbiomes (GMs) and epilepsy. However, the GM taxa identified in different studies are variable. In addition, observational studies cannot indicate causality. Therefore, our study aimed to explore the causal association of GMs with epilepsy and identify the most influential GM taxa. We conducted a Mendelian randomization (MR) study using summary statistics from genome-wide association studies (GWAS) of 211 GM taxa and epilepsy. The GWAS summary statistics for 211 GM taxa (from phylum to genus level) were generated by the MiBioGen consortium, while the FinnGen consortium provided the GWAS summary statistics for epilepsy. The primary analytical method to assess causality was the inverse-variance weighted (IVW) approach. To complement the IVW method, we also applied four additional MR methods: MR-Egger, weighted median, simple mode, and weighted. In addition, we conducted sensitivity analyses using Cochrane's Q-test, MR-Egger intercept test, MR-PRESSO global test, and leave-one-out analysis. We evaluated the causal effect of 211 GM taxa (from phylum to genus level) on epilepsy, generalized epilepsy, and focal epilepsy. After using the Bonferroni method for multiple testing correction, Class Betaproteobacteria [odds ratio (OR) = 1.357, 95% confidence interval (CI): 1.126-1.635, p = 0.001] and Order Burkholderiales (OR = 1.336, 95% CI: 1.112-1.606, p = 0.002). In addition, 21 nominally significant causal relationships were also identified. Further, the MR-Egger intercept test and MR-PRESSO global test suggested that our MR analysis was unaffected by horizontal pleiotropy (p > 0.05). Finally, the leave-one-out analysis suggested the robustness of the results. Through the MR study, we analyzed the causal relationship of 211 GM taxa with epilepsy and determined the specific intestinal flora associated with increased epilepsy risk. Our findings may provide helpful biomarkers for disease progression and potential candidate therapeutic targets for epilepsy. In addition, in-depth analysis of large-scale microbiome GWAS datasets based on metagenomics sequencing is necessary for future studies.

Multiple Hypothesis Testing in Conjoint Analysis

AbstractConjoint analysis is widely used for estimating the effects of a large number of treatments on multidimensional decision-making. However, it is this substantive advantage that leads to a statistically undesirable property, multiple hypothesis testing. Existing applications of conjoint analysis except for a few do not correct for the number of hypotheses to be tested, and empirical guidance on the choice of multiple testing correction methods has not been provided. This paper first shows that even when none of the treatments has any effect, the standard analysis pipeline produces at least one statistically significant estimate of average marginal component effects in more than 90% of experimental trials. Then, we conduct a simulation study to compare three well-known methods for multiple testing correction, the Bonferroni correction, the Benjamini–Hochberg procedure, and the adaptive shrinkage (Ash). All three methods are more accurate in recovering the truth than the conventional analysis without correction. Moreover, the Ash method outperforms in avoiding false negatives, while reducing false positives similarly to the other methods. Finally, we show how conclusions drawn from empirical analysis may differ with and without correction by reanalyzing applications on public attitudes toward immigration and partner countries of trade agreements.

Clinical and Radiographic Outcomes after Total Ankle Arthroplasty and Suture Tape Augmentation for Ankle Arthritis Combined with Ankle Instability

Category: Ankle Arthritis; Ankle; Sports Introduction/Purpose: End-stage ankle arthritis is commonly seen in combination with various forms of ligamentous instability. Correction of the ankle arthritis and meticulous ligament balancing is key to a successful outcome after a total ankle replacement. In particular, with varus or multidirectional instability, ligament repair or reconstruction can be performed with a combination of suture and/or suture anchors if there is adequate biologic tissue remaining for repair. To date there has been no study to report on the isolated use of suture tape to correct residual intraoperative lateral ankle instability and talar tilt after total ankle replacement. The purpose of this study is to evaluate the clinical and radiographic outcomes after this combined surgical procedure. Methods: This was a single institution, retrospective study involving 36 patients that underwent a primary total ankle arthroplasty surgery in combination with a lateral ankle ligament reconstruction using a suture tape construct (InternalBraceTM, Arthrex, Naples, FL). Retrospective review of prospectively collected information from patient charts of three foot and ankle fellowship trained surgeons were evaluated at preoperative, 6-months, 1-year and 2-years postoperative for various demographics as well as VAS pain and FAAM scores. Postoperative complications including revision surgery, hardware failure, or persistent instability were also recorded. In addition, preoperative radiographs were compared to 1-year and final follow up (mean, 26 months; range, 13-90 months) radiographs for correction of the coronal deformity. Analysis involved a 1-way repeated Friedman test with post hoc Wilcoxon signed-rank tests. P values are adjusted using the Bonferroni multiple testing correction method. Assumptions for normality were checked through normal quantile-quantile (QQ) plots. Results: At two years postoperative, VAS pain (p<0.001) and FAAM functional scores (p<0.001) were significantly improved from preoperative. Additionally, both VAS pain (p=0.026, p-0.043) and FAAM scores (p=0.00142, p<0.001) significantly improved at each time point (6-months, 1-year) compared to preoperative. On weight bearing mortise view radiographs, the mean preoperative talar tilt angle was 11.9 (range, 1.4-25.7) degrees of varus. At 1-year and 2-years postoperative the mean talar tilt angles were significantly improved from preoperative at 1.1 and 0.7 degrees, respectively (p<0.001, p<0.001)). The correction was significantly improved and maintained at final follow up of at least 2 years. There were no significant differences between 1-year and 2-year talar tilt measurements (p=1.00). There were three complications requiring revision surgery. Two were for component loosening and the third was for a deep infection. Conclusion: Combined ankle arthritis and instability can be a challenging problem for foot and ankle surgeons. This study demonstrated that the use of suture tape to reconstruct the lateral ankle ligament complex after total ankle replacement provides and maintains adequate correction of any residual coronal deformity at a final follow up of at least two years.

The Epithelial‐Mesenchymal Transition and Survival in Chromophobe Renal Cell Carcinoma

Introduction and ObjectiveAdvances in our understanding of the epithelial‐mesenchymal transition (EMT), how cells change polarity and surface‐expressed proteins in acquiring a more malignant state, has aided scientists in understanding cancer progression. Using in silico tools, this study examines how EMT‐related gene expression levels are associated with survival outcomes in the rare chromophobe renal cell carcinoma (ChRCC).MethodsUsing dbEMT, a curated database examining EMT literature, 1184 genes were selected as “EMT‐related” for further analysis. Kaplain‐Meier Survival plots were generated based on TCGA data on all recorded ChRCC patients for each gene (n=32 in “high” and “low” groups each based on a median cut‐off), with the step‐up Bonferroni method for multiple hypothesis testing correction being used (significance being established at p<0.004). Genes identified to be statistically significant in terms of survival outcome were selected for enrichment analysis through Cytospace to examine which biological pathways might predict survival outcome. Network analysis based on physical and predicted interactions using GeneMANIA and R was done to generate a list of non‐EMT related genes which might act with those statistically significant EMT‐related genes to affect survival, with cut‐off for for significance of both network and enrichment analysis being p<0.01. DisGeNET was used to observe if the gene expression profile of ChRCC resembled any other kind of cancer.ResultsFrom the original 1184 genes, 13 showed significant change following multiple hypothesis testing correction (p<0.0038), with all showing an increased expression relating to worse outcomes. Nine distinct processes, ranging from histone modification to SMAD2/3 nuclear pathways among others, were significantly overrepresented (p<0.01) in those 13 genes. The significant EMT‐related gene signature in ChRCC mimicked the profile of mantle cell lymphoma and urothelial carcinoma. Network analysis highlighted more than 20 additional genes which EMT‐related genes could interact with to affect survival.ConclusionsThis study showcases a small subset of EMT‐related genes significantly overexpressed in ChRCC cases with worse prognosis. Analysis highlighted pathway dysfunction regarding microtubule cytoskeletal organization and histone modification (among others) as a possible cause of increased mortality in cases of chromophobe renal cell carcinoma. Such a preliminary in silico analysis would allow for strategically targeted wet‐lab studies.

Meta-analysis of gene expression data in adipose tissue reveals new obesity associated genes.

High-throughput transcriptomic and proteomic data like microarray data are deposited in public databases such as Gene Expression Omnibus (GEO). Omics data integration and processing from different and independent studies is achieved by using efficient and effective computational tools through meta-analysis. Meta-analysis is a statistical powerful tool combining data from numerous studies, minimizes bias and increases statistical power by increasing sample size compared to individual studies. Therefore, we performed a meta-analysis of gene expression data in adipose tissue to identify genes that are differentially expressed between obese and non-obese subjects as well as to detect gene expression signatures, pathways and networks associated with obesity. We identified 821 differentially expressed genes (DEGs) in adipose tissue of obese subjects compared to non-obese. A protein-protein interactions (PPIs) network was reconstructed consisting of 168 proteins. Functional enrichment analysis in the network revealed proteins involved in RNA and energy metabolism. The KEGG pathway analysis revealed 15 enriched pathway terms. Furthermore, multiple testing correction methods identified five statistically significant obesity associated genes (NDUFA12, SFI1, SSB, FAR2 and LACE1) that require further investigation.

Development of a bioinformatics platform for analysis of quantitative transcriptomics and proteomics data: the OMnalysis.

BackgroundIn the past decade, RNA sequencing and mass spectrometry based quantitative approaches are being used commonly to identify the differentially expressed biomarkers in different biological conditions. Data generated from these approaches come in different sizes (e.g., count matrix, normalized list of differentially expressed biomarkers, etc.) and shapes (e.g., sequences, spectral data, etc.). The list of differentially expressed biomarkers is used for functional interpretation and retrieve biological meaning, however, it requires moderate computational skills. Thus, researchers with no programming expertise find difficulty in data interpretation. Several bioinformatics tools are available to analyze such data; however, they are less flexible for performing the multiple steps of visualization and functional interpretation.ImplementationWe developed an easy-to-use Shiny based web application (named as OMnalysis) that provides users with a single platform to analyze and visualize the differentially expressed data. The OMnalysis accepts the data in tabular form from edgeR, DESeq2, MaxQuant Perseus, R packages, and other similar software, which typically contains the list of differentially expressed genes or proteins, log of the fold change, log of the count per million, the P value, q-value, etc. The key features of the OMnalysis are multiple image type visualization and their dimension customization options, seven multiple hypothesis testing correction methods to get more significant gene ontology, network topology-based pathway analysis, and multiple databases support (KEGG, Reactome, PANTHER, biocarta, NCI-Nature Pathway Interaction Database PharmGKB and STRINGdb) for extensive pathway enrichment analysis. OMnalysis also fetches the literature information from PubMed to provide supportive evidence to the biomarkers identified in the analysis. In a nutshell, we present the OMnalysis as a well-organized user interface, supported by peer-reviewed R packages with updated databases for quick interpretation of the differential transcriptomics and proteomics data to biological meaning.AvailabilityThe OMnalysis codes are entirely written in R language and freely available at https://github.com/Punit201016/OMnalysis. OMnalysis can also be accessed from - http://lbmi.uvlf.sk/omnalysis.html. OMnalysis is hosted on a Shiny server at https://omnalysis.shinyapps.io/OMnalysis/. The minimum system requirements are: 4 gigabytes of RAM, i3 processor (or equivalent). It is compatible with any operating system (windows, Linux or Mac). The OMnalysis is heavily tested on Chrome web browsers; thus, Chrome is the preferred browser. OMnalysis works on Firefox and Safari.

Optimizing expression quantitative trait locus mapping workflows for single-cell studies

BackgroundSingle-cell RNA sequencing (scRNA-seq) has enabled the unbiased, high-throughput quantification of gene expression specific to cell types and states. With the cost of scRNA-seq decreasing and techniques for sample multiplexing improving, population-scale scRNA-seq, and thus single-cell expression quantitative trait locus (sc-eQTL) mapping, is increasingly feasible. Mapping of sc-eQTL provides additional resolution to study the regulatory role of common genetic variants on gene expression across a plethora of cell types and states and promises to improve our understanding of genetic regulation across tissues in both health and disease.ResultsWhile previously established methods for bulk eQTL mapping can, in principle, be applied to sc-eQTL mapping, there are a number of open questions about how best to process scRNA-seq data and adapt bulk methods to optimize sc-eQTL mapping. Here, we evaluate the role of different normalization and aggregation strategies, covariate adjustment techniques, and multiple testing correction methods to establish best practice guidelines. We use both real and simulated datasets across single-cell technologies to systematically assess the impact of these different statistical approaches.ConclusionWe provide recommendations for future single-cell eQTL studies that can yield up to twice as many eQTL discoveries as default approaches ported from bulk studies.

Abstract LB-174: Genome-wide DNA methylation signatures predict the early asymptomatic doxorubicin-induced cardiotoxicity in breast cancer

Abstract Background: Doxorubicin (DOX)-induced cardiotoxicity is cumulative-dose-dependent that begins with the first dose, but epigenetic changes after the first dose have not been examined. This study had two aims: Aim 1: to examine whether DNA methylation profile of peripheral blood cells (PBCs) induced by the first cycle of DOX-based chemotherapy can predict the risk of cardiotoxicity; and Aim 2 to determine if there are pretreatment methylation signatures at baseline to predict the risk of cardiotoxicity. Methods: Whole-genome DNA methylation of PBCs from 19 breast cancer patients was examined prior to the start of DOX-based chemotherapy and after the first cycle using Infinium HumanMethylation 450 BeadChip. ChAMP R package was used for the data normalization by the BMIQ algorithm and the COMBAT algorithm to adjust for batch effects. Chemotherapy-induced cardiotoxicity was determined by the assessment of left ventricle ejection fraction (LVEF) with multigated acquisition (MUGA) scan before chemotherapy and at its completion, and any LVEF decline by >10% was considered abnormal. Results: Pre- and post- DOX treatment were compared between patients with abnormal LVEF decline and those who maintained normal LVEF to assess the differences in the cytosine methylation states. The multiple correction testing resulted in a total of 515 differentially methylated CpGs between the patients with abnormal versus normal LVEF post-DOX treatment. There were 379 differentially methylated CpGs at baseline and 136 CpGs after the first cycle which significantly correlated with LVEF status after applying the Holm-Bonferroni method for multiple testing correction. The intersection of the probes resulted in 38 differentially methylated genes (DMG) that when used in hierarchical clustering, correctly clustered baseline samples into those that would result in normal and abnormal LVEF after treatment. DMG genes included SEPT5, IRF6, KCNQ1, NELL1, SIX6, HOXC9 and BCAN. IRF6, which encodes interferon regulatory factor 6, was identified as being significant in the subsequent analysis at the probe and region level. Comparisons of the samples before and after DOX treatment did not result in any probes associated with the treatment that were significant after multiple testing correction. Conclusions: The results from this study provide evidence that DNA methylation profile of peripheral blood cells has the potential to predict the risk of DOX-induced cardiotoxicity. The important finding was that the extent of methylation at baseline correlated with the post-DOX LVEF reduction, indicating that it may have the potential to predict the subsequent development of cardiotoxicity. Further studies with larger cohort of patients are needed to confirm these findings as well as narrowing down candidate methylation markers that can be implemented in a blood test. These finding may help guide treatment or identify patients that need to be followed closely to mitigate heart damage after DOX treatment. Citation Format: Michael A. Bauer, Issam Makhoul, Ping-Ching Hsu, Jeanne Wei, L. Joseph Su, Annjanette Stone, Weleetka Carter, Valentina K. Todorova. Genome-wide DNA methylation signatures predict the early asymptomatic doxorubicin-induced cardiotoxicity in breast cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-174.

Joint variable selection and network modeling for detecting eQTLs.

In this study, we conduct a comparison of three most recent statistical methods for joint variable selection and covariance estimation with application of detecting expression quantitative trait loci (eQTL) and gene network estimation, and introduce a new hierarchical Bayesian method to be included in the comparison. Unlike the traditional univariate regression approach in eQTL, all four methods correlate phenotypes and genotypes by multivariate regression models that incorporate the dependence information among phenotypes, and use Bayesian multiplicity adjustment to avoid multiple testing burdens raised by traditional multiple testing correction methods. We presented the performance of three methods (MSSL - Multivariate Spike and Slab Lasso, SSUR - Sparse Seemingly Unrelated Bayesian Regression, and OBFBF - Objective Bayes Fractional Bayes Factor), along with the proposed, JDAG (Joint estimation via a Gaussian Directed Acyclic Graph model) method through simulation experiments, and publicly available HapMap real data, taking asthma as an example. Compared with existing methods, JDAG identified networks with higher sensitivity and specificity under row-wise sparse settings. JDAG requires less execution in small-to-moderate dimensions, but is not currently applicable to high dimensional data. The eQTL analysis in asthma data showed a number of known gene regulations such as STARD3, IKZF3 and PGAP3, all reported in asthma studies. The code of the proposed method is freely available at GitHub (https://github.com/xuan-cao/Joint-estimation-for-eQTL).

DOP08 Deep remission in paediatric Crohn’s disease is associated with increased abundance of dialister species and increased valerate

Abstract Background Deep remission, or the absence of mucosal inflammation on endoscopy, is associated with improved long-term outcomes in paediatric Crohn’s disease (pCD). In pCD faecal calprotectin level <100 mg/l is a validated non-invasive marker of deep remission and is a therapeutic target used in clinical practice. The microbiome in pCD has been extensively compared with controls and is known to have reduced diversity, high instability, and depleted short-chain fatty acid (SCFA) producers. This study investigates variation in microbiome, metabolome and protease activity associated with deep remission. Methods Children aged 3–18 years with established pCD on infliximab were recruited. Stool samples were cooled immediately, aliquoted, and frozen at −80oC within 4 h. Deep remission status was assigned if the calprotectin level was under 100 mg/l. 16S sequencing and analysis used Illumina MiSeq (V1-2 primers), DADA2, and STAMP. ¹H-NMR spectral profiles were used to develop partial-least-squares discriminant-analysis with Monte-Carlo cross-validation (MCCV-PLS-DA) models. Metabolome microbiome correlation used Storey’s method for multiple testing correction and a threshold of │RHO│≥ 0.5 with q < 0.05. Protease activity was assayed by comparing the breakdown of fluorescein labelled casein against trypsin standards. Results Twenty-one children were included; 10 were in deep remission and 11 had active disease. There were no significant differences in age, disease duration, treatment, or BMI between groups. Mean height z score was lower in those with active disease (p = 0.015). There was no significant difference in α diversity between groups, and non-metric multidimensional scaling of weighted unifrac distance identified no difference in community composition (PERMANOVA p = 0.317). The relative abundance of Dialister species was higher in those with deep remission (3.4 vs. 0.4%, q = 0.033). Dialister abundance was positively correlated with valerate concentration. MCCV–PLS-DA was able to effectively discriminate metabolome profiles from active and deep remission patients (R2 = 0.87, Q2 = 0.51); valerate and glutamate were increased in deep remission, ethanol and lysine were decreased. There was a trend towards increased protease activity in patients with active disease (p = 0.054). Conclusion Dialister species are enriched in pCD in deep remission and are associated with significant metabolome changes, including increased valerate. Valerate, like other SCFAs, has known anti-inflammatory properties and has previously been implicated in effective C. difficile treatment. This study supports an association between Dialister abundance, valerate and mucosal healing in pCD. A trend towards higher protease activity in active disease also warrants further investigation.

GenFam: A web application and database for gene family-based classification and functional enrichment analysis.

Genome‐scale studies using high‐throughput sequencing (HTS) technologies generate substantial lists of differentially expressed genes under different experimental conditions. These gene lists need to be further mined to narrow down biologically relevant genes and associated functions in order to guide downstream functional genetic analyses. A popular approach is to determine statistically overrepresented genes in a user‐defined list through enrichment analysis tools, which rely on functional annotations of genes based on Gene Ontology (GO) terms. Here, we propose a new computational approach, GenFam, which allows annotation, classification, and enrichment of genes based on their gene family, thus simplifying identification of candidate gene families and associated genes that may be relevant to the query. GenFam and its integrated database comprises of three hundred and eighty‐four unique gene families and supports gene family analyses for sixty plant genomes. Four comparative case studies with plant species belonging to different clades and families were performed using GenFam which demonstrated its robustness and comprehensiveness over preexisting functional enrichment tools. To make it readily accessible for plant biologists, GenFam is available as a web‐based application where users can input gene IDs and export enrichment results in both tabular and graphical formats. Users can also customize analysis parameters by choosing from the various statistical enrichment tests and multiple testing correction methods. Additionally, the web‐based application, source code, and database are freely available to use and download. Website: http://mandadilab.webfactional.com/home/. Source code and database: http://mandadilab.webfactional.com/home/dload/.

Power, false discovery rate and Winner's Curse in eQTL studies.

Investigation of the genetic architecture of gene expression traits has aided interpretation of disease and trait-associated genetic variants; however, key aspects of expression quantitative trait loci (eQTL) study design and analysis remain understudied. We used extensive, empirically driven simulations to explore eQTL study design and the performance of various analysis strategies. Across multiple testing correction methods, false discoveries of genes with eQTLs (eGenes) were substantially inflated when false discovery rate (FDR) control was applied to all tests and only appropriately controlled using hierarchical procedures. All multiple testing correction procedures had low power and inflated FDR for eGenes whose causal SNPs had small allele frequencies using small sample sizes (e.g. frequency <10% in 100 samples), indicating that even moderately low frequency eQTL SNPs (eSNPs) in these studies are enriched for false discoveries. In scenarios with ≥80% power, the top eSNP was the true simulated eSNP 90% of the time, but substantially less frequently for very common eSNPs (minor allele frequencies >25%). Overestimation of eQTL effect sizes, so-called ‘Winner’s Curse’, was common in low and moderate power settings. To address this, we developed a bootstrap method (BootstrapQTL) that led to more accurate effect size estimation. These insights provide a foundation for future eQTL studies, especially those with sampling constraints and subtly different conditions.

Abstract A18: Metabolomic analysis of ovarian cancer risk in the Nurses’ Health Studies: Metabolite associations are more pronounced in non-serous tumors

Abstract Background: To identify potential novel risk factors for ovarian cancer, we assessed the role of metabolites and lipids in the development of ovarian cancer. Prior research suggests that lipid synthesis and metabolism are dysregulated in ovarian tumors. Furthermore, studies show that lysophosphatidylcholines (LPC), PCs, and sphingomyelins (SM) are differentially expressed in ovarian cancer cases versus controls. We used a validated metabolomics platform to assess prediagnosis measures of these lipid classes with ovarian cancer risk and conduct a discovery-based analysis of other metabolites. Methods: LC-MS/MS targeted metabolomics were measured at the Broad Institute (Cambridge, MA, USA). 300 cases were matched to one control on age, menopause status/hormone therapy use, fasting, season, and time of day of blood draw. 348 metabolites and lipids passed QC. Metabolites were transformed using probit scores for normality. We used conditional logistic regression to identify metabolites and lipids associated with total invasive ovarian cancer risk. We used unconditional logistic regression, adjusting for matching factors, in histotype-specific analyses including all available controls, thus increasing power. All analyses adjusted for known risk factors: oral contraceptive use, parity, and tubal ligation. Results: We found 18 metabolites that were associated with risk of invasive ovarian cancer (p≤0.05). These included sphingosine, C3-DC-CH3 carnitine, choline, cytosine, triacylglycerols (TAG), C18:1 SM, lysophosphatidylinositol (LPI), mono hydroxyoctadecadienoic acid, phosphatidylethanolamine (PE), and taurine. The odds ratios ranged between 0.73 and 0.84 per 1 standard deviation (SD) increase in metabolite value. None of the metabolites remained significant after multiple testing correction (false discovery rate (FDR) corrected p≥0.85). In histotype analysis, 13 metabolites were associated with serous disease (p≤0.05). These included C3-DC-CH3 carnitine, 7 TAGs, choline, symmetric dimethylarginine (SDMA), citrulline, cytosine, and 6-deoxodolichosterone. Odds ratios ranged between 0.75 and 0.82 for 1 SD increase in the metabolite values. No metabolites remained significant after multiple testing correction (FDR-corrected p≥0.75). Interestingly, 40 metabolites were associated with non-serous ovarian cancer at p≤0.05. Nearly all were different than those found for total invasive or serous disease. These included pregnanediol-3-glucuronide, laserpitin, trihydroxyecdysone, docosahexaenoic acid, decosapentaenoic acid, niacinamide, testosterone sulfate, alanine, taurine, eicosapentaenoic acid, thiamine, 1-methyladenosine, cajaisoflavone, kurilensoside H, and multiple lipid classes (cholesterol ester (CE), PCs, PE and PC plasmalogens, TAGs, carnitines, SMs, DAG, LPI, lysophosphatidic acid (LPA), lysophosphatidylethanolamine (LPE), PE, TAGs, and diacylglycerols (DAGs)). Odds ratios for protective metabolites ranged between 0.49 and 0.74 while for harmful metabolites they ranged between 1.38 and 1.48 for 1 SD increase in the metabolite values. However, FDR-corrected p-values were ≥0.41. Conclusion: While several metabolites differed between ovarian cancer cases and controls on a nominal p-value scale, none remained significant after multiple testing correction. However, this test does not account for the high correlation between many of the metabolites. Notably, distinct markers were associated with different histotypes and more metabolites were associated with non-serous tumors, which have been associated more strongly with BMI and other metabolic risk factors. For nonserous disease, we observed associations with LPCs, PCs, and SM lipids, which were part of our a priori hypotheses, as well as sex hormone metabolites. Ongoing analyses include adjusting for BMI at age 18 and adolescent somatotype to account for early life adiposity, consideration of other multiple testing correction methods, and pathway analyses. Citation Format: Oana A. Zeleznik, Elizabeth M. Poole, Clary Clish, Heather A. Eliassen, Peter Kraft, Shelley S. Tworoger. Metabolomic analysis of ovarian cancer risk in the Nurses’ Health Studies: Metabolite associations are more pronounced in non-serous tumors. [abstract]. In: Proceedings of the AACR Conference: Addressing Critical Questions in Ovarian Cancer Research and Treatment; Oct 1-4, 2017; Pittsburgh, PA. Philadelphia (PA): AACR; Clin Cancer Res 2018;24(15_Suppl):Abstract nr A18.

Myriads: P-value-based multiple testing correction.

There are many multiple testing correction methods. Some of them are robust to various dependencies in the data while others are not. Some of the implementations have problems for managing high dimensional list of P-values as currently demanded by microarray and other omic technologies. The program Myriads, formerly SGoF+, provides some of the most important P-value-based correction methods jointly with a test of dependency and a P-value simulator. Myriads easily manage hundreds of thousands of P-values. http://myriads.webs.uvigo.es. myriads@uvigo.es. Supplementary data are available at Bioinformatics online.

Role of DNA methylation at the placental RTL1 gene locus in type 1 diabetes.

Genome-wide association studies (GWAS) have identified more than 40 T1D loci associated with type 1 diabetes (T1D). How these polymorphisms interact with environmental factors to trigger T1D is unknown, but recent evidence suggests that epigenetic mechanisms could play a role. To begin to explore the contribution of epigenetics to T1D, we have examined DNA methylation in a pilot study of whole blood cells DNA from 10 young T1D patients and 10 young controls. Through the study of >900 000 CG loci across a diverse set of functionally relevant genomic regions using a custom DNA methylation array, we identified 250 T1D-differentially methylated region (DMR) at p < 0.05 and 1 DMR using next a permutation-based multiple testing correction method. This DMR is located in an imprinted region previously associated with T1D on the chromosome 14 that encompasses RTL1 gene and 2 miRNAs (miR136 and miR432). Using pyrosequencing-based bisulfite PCR, we replicated this association in a different and larger set of T1D patients and controls. DNA methylation at this DMR was inversely correlated with RTL1 gene expression and positively correlated with miR136 expression in human placentas. The DMR identified in this study presents suggestive evidence for altered methylation site in T1D and provide a promising new candidate gene. RTL1 is essential for placental permeability function in the mid-to-late fetal stages. We suggest that hypo-methylation could increase the fetal exposure to environmental factors in T1D susceptibility.

Gene Prioritization for Imaging Genetics Studies Using Gene Ontology and a Stratified False Discovery Rate Approach.

Imaging genetics is an emerging field in which the association between genes and neuroimaging-based quantitative phenotypes are used to explore the functional role of genes in neuroanatomy and neurophysiology in the context of healthy function and neuropsychiatric disorders. The main obstacle for researchers in the field is the high dimensionality of the data in both the imaging phenotypes and the genetic variants commonly typed. In this article, we develop a novel method that utilizes Gene Ontology, an online database, to select and prioritize certain genes, employing a stratified false discovery rate (sFDR) approach to investigate their associations with imaging phenotypes. sFDR has the potential to increase power in genome wide association studies (GWAS), and is quickly gaining traction as a method for multiple testing correction. Our novel approach addresses both the pressing need in genetic research to move beyond candidate gene studies, while not being overburdened with a loss of power due to multiple testing. As an example of our methodology, we perform a GWAS of hippocampal volume using both the Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA2) and the Alzheimer's Disease Neuroimaging Initiative datasets. The analysis of ENIGMA2 data yielded a set of SNPs with sFDR values between 10 and 20%. Our approach demonstrates a potential method to prioritize genes based on biological systems impaired in a disease.

Abstract B35: Cumulative results from an investigation of transcriptome differences in breast cancer samples from African American and East African triple-negative breast cancer patients

Abstract We report our cumulative results of a multi-year study focusing on the genomic similarities and differences between Triple-negative breast cancer (TNBC) in native African and U.S. African patients. We used a cohort of African-American (AA) TNBC patients from South Florida and an indigenous cohort of Kenyan TNBC patients. Strong U.S. epidemiological data supports breast cancer (BC) as the second leading cause of cancer death among AA women, with a 20% greater mortality rate than that in Caucasian-American women (CA). AA women often present with tumors that are of higher grade and later stage and their relapse-free and overall survival period is significantly lower than CA. Additionally, TNBC has a higher prevalence in AA women. All of these observations lead to a much poorer prognosis for AA BC patients- thus defining the all-encompassing term “disparities in BC in women of African ancestry”. In our study, we used archived formalin fixed paraffin embedded (FFPE) blocks, collected across a stage-matched multi-ethnic U.S. cohort of node-negative TNBC from the University of Miami and similar TNBC samples from the Kijabe Hospital (Kenya). Using 10 μm scrolls from each block, total RNA was isolated, cDNA prepared, and hybridized to a breast cancer enriched gene expression array (Affymetrix Platform BC DSA Research Tool) in collaboration with Almac Diagnostics. Gene expression analysis was conducted using GeneSpring 12.1® analytical software. The raw data, a total of 60,856 gene/probes, was pre-processed using Robust MultiArray Average (RMA) algorithm; background corrected, log2 transformed to the baseline median of all samples and then quintile normalized. The samples that passed final QC were then filtered to remove low-intensity signals (&amp;lt; 20% signal). The final analysis cohort consisted of 10-AA, 13-CA and 21-Kenyan. PCA analysis revealed that the samples clustered well with respect to ethnicity. Unsupervised cluster analysis, based on ethnicity and genes (p value &amp;lt;.05, fold change &amp;gt;2.5), was performed. The resulting dendogram clearly segregated into distinct subgroups based on ethnicity, revealing a pattern of differential gene expression between the cohorts. A list of differentially expressed genes from each cohort (DEG) were selected using ANOVA analysis (fold change &amp;gt; 3.0, p value &amp;lt;.05) followed by the Benjamin and Hochberg method for multiple-testing correction. Finally, the lists of DEG were uploaded into GeneGo MetaCore to identify functionally enriched pathways. These analyses revealed significantly deregulated genes associated with the Wnt/β-catenin pathway in the AA cohort, as compared to the CA, suggesting that this pathway may contribute to the more aggressive phenotype in AA women diagnosed with TNBC. Additionally, significantly deregulated genes associated with the Oncostatin M pathway were discovered in the Kenyan cohort, as compared to the AA and CA tumors. In particular, STAT1 was significantly downregulated in the Kenyan cohort compared to the CA and AA. Thus, our results indicate gene expression differences within several key pathways across these ethnic groups. In summary, this study represents the first direct comparison of gene expression in TNBC specimens across U.S. CA and AA BC patients and Kenyan East Africans. These studies have important implications for further understanding BC ethnic disparities, as well as tailored approaches to prediction, prevention and therapeutic measures. Citation Format: Julie E. Getz, Mary E. Ahearn, Carmen Gomez, Biju Issac, Jennifer Clarke, Mark D. Pegram, Peter Bird, John D. Carpten, Lisa L. Baumbach. Cumulative results from an investigation of transcriptome differences in breast cancer samples from African American and East African triple-negative breast cancer patients. [abstract]. In: Proceedings of the Sixth AACR Conference: The Science of Cancer Health Disparities; Dec 6–9, 2013; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2014;23(11 Suppl):Abstract nr B35. doi:10.1158/1538-7755.DISP13-B35