Adaptive False Discovery Rate Control for Heterogeneous Data

Recent advances in long-read sequencing technologies have empowered the detection of structural variations (SVs) associated with genetic diseases. Despite the availability of numerous SV callers and efforts to merge SVs from multiple tools, there remains limited research on quantifying the confidence levels for the reported results. In this work, we propose a Bayesian integration model that combines SV calls from different tools. Notably, we introduce an approach for false discovery rate (FDR) control and provide a quantitative measure for the merged SVs. Our model can handle cases where certain tools lack quality scores, showcasing its flexibility in incorporating additional tools. Through extensive simulation studies, we evaluate the performance of our method under various conditions, demonstrating the FDR estimation accuracy and improved F1 score. Furthermore, we validate our model using simulated human genome sequencing data and the HG002dataset.

The multiplicity problem has become increasingly important in genetic studies as the capacity for high-throughput genotyping has increased. The control of False Discovery Rate (FDR) (Benjamini and Hochberg. [1995] J. R. Stat. Soc. Ser. B 57:289-300) has been adopted to address the problems of false positive control and low power inherent in high-volume genome-wide linkage and association studies. In many genetic studies, there is often a natural stratification of the m hypotheses to be tested. Given the FDR framework and the presence of such stratification, we investigate the performance of a stratified false discovery control approach (i.e. control or estimate FDR separately for each stratum) and compare it to the aggregated method (i.e. consider all hypotheses in a single stratum). Under the fixed rejection region framework (i.e. reject all hypotheses with unadjusted p-values less than a pre-specified level and then estimate FDR), we demonstrate that the aggregated FDR is a weighted average of the stratum-specific FDRs. Under the fixed FDR framework (i.e. reject as many hypotheses as possible and meanwhile control FDR at a pre-specified level), we specify a condition necessary for the expected total number of true positives under the stratified FDR method to be equal to or greater than that obtained from the aggregated FDR method. Application to a recent Genome-Wide Association (GWA) study by Maraganore et al. ([2005] Am. J. Hum. Genet. 77:685-693) illustrates the potential advantages of control or estimation of FDR by stratum. Our analyses also show that controlling FDR at a low rate, e.g. 5% or 10%, may not be feasible for some GWA studies.

The methodological advancement in microarray data analysis on the basis of false discovery rate (FDR) control, such as the q-value plots, allows the investigator to examine the FDR from several perspectives. However, when FDR control at the "customary" levels 0.01, 0.05, or 0.1 does not provide fruitful findings, there is little guidance for making the trade off between the significance threshold and the FDR level by sound statistical or biological considerations. Thus, meaningful statistical significance criteria that complement the existing FDR methods for large-scale multiple tests are desirable. Three statistical significance criteria, the profile information criterion, the total error proportion, and the guide-gene driven selection, are developed in this research. The first two are general significance threshold criteria for large-scale multiple tests; the profile information criterion is related to the recent theoretical studies of the connection between FDR control and minimax estimation, and the total error proportion is closely related to the asymptotic properties of FDR control in terms of the total error risk. The guide-gene driven selection is an approach to combining statistical significance and the existing biological knowledge of the study at hand. Error properties of these criteria are investigated theoretically and by simulation. The proposed methods are illustrated and compared using an example of genomic screening for novel Arf gene targets. Operating characteristics of q-value and the proposed significance threshold criteria are investigated and compared in a simulation study that employs a model mimicking a gene regulatory pathway. A guideline for using these criteria is provided. Splus/R code is available from the corresponding author upon request.

Olfaction dysfunction is considered as a robust marker of prodromal Parkinson disease (PD). Measurement of olfaction function as a screening test is unsatisfactory due to long lead time interval and low specificity for detection of PD. Use of imaging markers might yield more accurate predictive values and provide bases for combined use of imaging and clinical markers for early PD. Diffusion MRI connectometry was conducted on 85 de novo PD patients in and 36 healthy controls to find: first, white matter tracts with significant difference in quantitative anisotropy between PD groups with various degrees of olfaction dysfunction and second, second fibers with correlation with University of Pennsylvania Smell Identification Test (UPSIT) score in each group using a multiple regression analysis considering age, sex, GDS and MoCA score. Local connectomes were determined in seven of all the possible comparisons, correcting for false discovery rate (FDR). PD patients with anosmia and normal olfaction had the highest number of fibers with decreased connectivity in left inferior longitudinal fasciculus, bilateral fornix, bilateral middle cerebellar peduncle (MCP), bilateral cingulum, bilateral corticospinal tract (CST) and body, genu and splenium of corpus callosum (CC) (FDR = 0.0013). In multiple regression analysis, connectivity in the body, genu and splenium of CC and bilateral fornix had significant negative correlation (FDR between 0.019 and 0.083), and bilateral cingulum and MCP had significant positive correlation (FDR between 0.022 and 0.092) with UPSIT score. White matter connectivity in healthy controls could not be predicted by UPSIT score using the same model. The results of this study provide compelling evidence that microstructural degenerative changes in these areas underlie the clinical phenotype of prodromal olfaction dysfunction in PD and that diffusion parameters of these areas might be able to serve as signature markers for early detection of PD. This is the first report that confirms a discriminative role for UPSIT score in identifying PD specific changes in white matter microstructure. Our results open a window to identify microstructural signatures of prodromal PD in white matter.

The Benjamini-Hochberg (BH) procedure is a celebrated method for multiple testing with false discovery rate (FDR) control. In this paper, we consider large-scale distributed networks where each node possesses a large number of p-values and the goal is to achieve the global BH performance in a communication-efficient manner. We propose that every node performs a local test with an adjusted test size according to the (estimated) global proportion of true null hypotheses. With suitable assumptions, our method is asymptotically equivalent to the global BH procedure. Motivated by this, we develop an algorithm for star networks where each node only needs to transmit an estimate of the (local) proportion of nulls and the (local) number of p-values to the center node; the center node then broadcasts a parameter (computed based on the global estimate and test size) to the local nodes. In the experiment section, we utilize existing estimators of the proportion of true nulls and consider various settings to evaluate the performance and robustness of our method.

The recent RNA-seq technology is an attractive method to study gene expression. One of the most important goals in RNA-seq data analysis is to detect genes differentially expressed across treatments. Although several statistical methods have been published, there are no theoretical justifications for whether these methods are optimal or how to search for the optimal test. Furthermore, most proposed tests are designed for testing whether the mean expression levels are exactly the same or not across treatments, whereas sometimes, biologists are interested in detecting genes with expression changes larger than a certain threshold. Another issue with current methods is that the false discovery rate (FDR) control is not well studied. In this manuscript, we propose a test to address all the above issues. Under model assumptions, we derive an optimal test that achieves the maximum of average power among those that control FDR at the same level. We also provide an approximated version, the approximated most average powerful (AMAP) test, for practical implementation. The proposed method allows for testing null hypotheses that are much more general than the ones most previous studies have considered, and it leads to a natural way of controlling the FDR. Through simulation studies, we show that our test has a higher power than other methods, including the widely-used edgeR, DESeq, and baySeq methods, as well as better FDR control than two other FDR control procedures commonly used in practice. For demonstration, we also apply the proposed method to a real RNA-seq dataset obtained from maize.

BackgroundIrritable bowel syndrome (IBS) is a common disease with unknown etiology. Poor dietary intake with nutritional deficiency and overweight have been described to increase the risk of IBS. The aim of the present study was to compare weight and circulating levels of micronutrients in IBS compared with healthy controls.DesignCross-sectional study.MethodsPatients diagnosed with IBS and healthy volunteers were recruited. Participants had to complete a dietary diary book and the questionnaires Rome IV, IBS-severity scoring system (IBS-SSS), and visual analog scale for IBS (VAS-IBS). Weight and height were measured, and blood samples were drawn. C-reactive protein (CRP), cobalamin, folate, iron, total iron-binding capacity (TIBC), and 25-hydroxy (25-OH) vitamin D were analyzed. Differences were calculated between groups and generalized linear model for regressions was adjusted for false discovery rate (FDR).ResultsIBS patients (n = 260) were elder than controls (n = 50) (44.00 (33.25-56.00) vs. 37.85 (30.18–45.48) years; p = 0.012). After adjustment for age, both weight (β: 5.880; 95% CI: 1.433–10.327; p = 0.010, FDR = 0.020) and body mass index (BMI) (β: 2.02; 95% CI: 0.68–3.36; p = 0.003, FDR = 0.012) were higher in patients. Among IBS participants, 48.1% were overweight/obese compared with 26.0% in controls (p = 0.007). Diarrhea-predominated IBS had highest weight (p < 0.001) and BMI (p = 0.077). CRP and cobalamin were higher in patients than controls (p = 0.010 vs. p = 0.007), whereas folate was highest in controls (p = 0.001). IBS patients had lower intake of vegetables (p = 0.026), dairy products (p = 0.004), and cereals (p = 0.010) compared with controls. Despite 21.5% of IBS patients were taking vitamin D supplements, 23.65% of them had vitamin D levels below 50 nmol/L, compared with 26.0% observed in the control group (p = 0.720). Vitamin D levels were lower in overweight than in normal weight IBS patients (60 (48–73) nmol/L vs. 65 (53–78) nmol/L, p = 0.022). Vitamin D correlated with cobalamin and folate but correlated inversely with TIBC and BMI. IBS patients had a high degree of gastrointestinal and extraintestinal symptoms, which were inversely associated with iron levels. Extraintestinal symptoms were associated with increased BMI.ConclusionIBS patients were often overweight or obese, with low vitamin D levels. High burden of extraintestinal symptoms were associated with overweight and lower iron levels.RegistrationClinicalTrials.gov, NCT05192603 (Date of registration 11/29/2021) and NCT03306381 (Date of registration 09/18/2017), respectively.

Signal detection in functional magnetic resonance imaging (fMRI) inherently involves the problem of testing a large number of hypotheses. A popular strategy to address this multiplicity is the control of the false discovery rate (FDR). In this work we consider the case where prior knowledge is available to partition the set of all hypotheses into disjoint subsets or families, e. g., by a-priori knowledge on the functionality of certain regions of interest. If the proportion of true null hypotheses differs between families, this structural information can be used to increase statistical power. We propose a two-stage multiple test procedure which first excludes those families from the analysis for which there is no strong evidence for containing true alternatives. We show control of the family-wise error rate at this first stage of testing. Then, at the second stage, we proceed to test the hypotheses within each non-excluded family and obtain asymptotic control of the FDR within each family at this second stage. Our main mathematical result is that this two-stage strategy implies asymptotic control of the FDR with respect to all hypotheses. In simulations we demonstrate the increased power of this new procedure in comparison with established procedures in situations with highly unbalanced families. Finally, we apply the proposed method to simulated and to real fMRI data.

False discovery rate (FDR) control serves as a critical safeguard in high-dimensional statistical analysis, increasing reliability and ensuring reproducibility of discoveries. Although extensively studied for real-valued data, FDR methodologies remain underdeveloped for the complex-valued data domain despite their importance in signal processing, physics, and engineering applications. This work confronts a central challenge in FDR-controlled analysis: settings in which the data contain groups of highly correlated variables. To achieve this, we advance our Complex-Valued Terminating-Random Experiments (CT-Rex) framework by developing the CT-Rex+GVS, a grouped variable selector. The method comes in two variants: the elastic net <tex>$(E N)$</tex> and the informed elastic net (IEN), both with isotropic and phase agnostic regularization. We validate the framework by benchmarking it in the selection of complex-valued grouped variables for linear regression and in the estimation of the singlesnapshot direction-of-arrival (DOA) based on compressed sensing using uniform linear arrays (ULA). The conducted numerical experiments confirm the FDR control property and demonstrate favorable performance compared to existing approaches.

Whole genome microarray investigations (e.g. differential expression, differential methylation, ChlP-Chip) provide opportunities to test millions of features in a genome. Traditional multiple comparison procedures such as familywise error rate (FWER) controlling procedures are too conservative. Although false discovery rate (FDR) procedures have been suggested as having greater power, the control itself is not exact and depends on the proportion of true null hypotheses. Because this proportion is unknown, it has to be accurately (small bias, small variance) estimated, preferably using a simple calculation that can be made accessible to the general scientific community. We propose an easy-to-implement method and make the R code available, for estimating the proportion of true null hypotheses. This estimate has relatively small bias and small variance as demonstrated by (simulated and real data) comparing it with four existing procedures. Although presented here in the context of microarrays, this estimate is applicable for many multiple comparison situations.

Discrete uniform and homogeneous p-values often arise in applications with multiple testing. For example, this occurs in genome wide association studies whenever a non-parametric one-sample (or two-sample) test is applied throughout the gene loci. In this paper, we consider multiple comparison procedures for such scenarios based on several existing estimators for the proportion of true null hypotheses, π0, which take the discreteness of the p-values into account. The theoretical guarantees of the several approaches with respect to the estimation of π0 and the false discovery rate control are reviewed. The performance of the discrete procedures is investigated through intensive Monte Carlo simulations considering both independent and dependent p-values. The methods are applied to three real data sets for illustration purposes too. Since the particular estimator of π0 used to compute the q-values may influence its performance, relative advantages and disadvantages of the reviewed procedures are discussed. Practical recommendations are given.

Proteomic Profiling Identifies Unique IDH Mutant Signatures and Novel Therapeutic Opportunities in Adult Acute Myeloid Leukemia Patients

Estimating the proportion of true null hypotheses, π0, has attracted much attention in the recent statistical literature. Besides its apparent relevance for a set of specific scientific hypotheses, an accurate estimate of this parameter is key for many multiple testing procedures. Most existing methods for estimating π0 in the literature are motivated from the independence assumption of test statistics, which is often not true in reality. Simulations indicate that most existing estimators in the presence of the dependence among test statistics can be poor, mainly due to the increase of variation in these estimators. In this paper, we propose several data-driven methods for estimating π0 by incorporating the distribution pattern of the observed p-values as a practical approach to address potential dependence among test statistics. Specifically, we use a linear fit to give a data-driven estimate for the proportion of true-null p-values in (λ, 1] over the whole range [0, 1] instead of using the expected proportion at 1−λ. We find that the proposed estimators may substantially decrease the variance of the estimated true null proportion and thus improve the overall performance.

In many multiple testing procedures, accurate modeling of the p-value distribution is a key issue. Mixture distributions have been shown to provide adequate models for p-value densities under the null and the alternative hypotheses. An important parameter of the mixture model that needs to be estimated is the proportion of true null hypotheses, which under the mixture formulation becomes the probability mass attached to the value associated with the null hypothesis. It is well known that in a general mixture model, especially when a scale parameter is present, the mixing distribution need not be identifiable. Nevertheless, under our setting for mixture model for p-values, we show that the weight attached to the null hypothesis is identifiable under two very different types of conditions. We consider several examples including univariate and multivariate mixture models for transformed p-values. Finally, we formulate an abstract theorem for general mixtures and present other examples.

A topic which is becoming more and more popular in Functional Data Analysis is local inference, i.e., the continuous statistical testing of a null hypothesis along a domain of interest. The principal issue in this topic is the infinite amount of tested hypotheses, which can be seen as an extreme case of multiple comparisons problem.Anumber of quantities have been introduced in the literature of multivariate analysis in relation to the multiple comparisons problem. Arguably the most popular one is the False Discovery Rate (FDR), that measures the expected proportion of false discoveries (rejected null hypotheses) among all discoveries.We define FDR in the setting of functional data defined on a compact set of Rd. A continuous version of the Benjamini-Hochberg procedure is introduced, along with a definition of adjusted 𝑝-value function. Some general conditions are stated, under which the functional Benjamini-Hochberg (fBH) procedure provides control of FDR. We show how the procedure can be plugged-in with every parametric or nonparametric pointwise test, given that such test is exact. Finally, the proposed method - together with a nonparametric test - is applied to the analysis of the benchmark dataset of Canadian temperatures.