Landscape of Conditional eQTL in Dorsolateral Prefrontal Cortex and Co-localization with Schizophrenia GWAS.

Causal genes and variants within genome-wide association study (GWAS) loci can be identified by integrating GWAS statistics with expression quantitative trait loci (eQTL) and determining which variants underlie both GWAS and eQTL signals. Most analyses, however, consider only the marginal eQTL signal, rather than dissect this signal into multiple conditionally independent signals for each gene. Here we show that analyzing conditional eQTL signatures, which could be important under specific cellular or temporal contexts, leads to improved fine mapping of GWAS associations. Using genotypes and gene expression levels from post-mortem human brain samples (n = 467) reported by the CommonMind Consortium (CMC), we find that conditional eQTL are widespread; 63% of genes with primary eQTL also have conditional eQTL. In addition, genomic features associated with conditional eQTL are consistent with context-specific (e.g., tissue-, cell type-, or developmental time point-specific) regulation of gene expression. Integrating the 2014 Psychiatric Genomics Consortium schizophrenia (SCZ) GWAS and CMC primary and conditional eQTL data reveals 40 loci with strong evidence for co-localization (posterior probability > 0.8), including six loci with co-localization of conditional eQTL. Our co-localization analyses support previously reported genes, identify novel genes associated with schizophrenia risk, and provide specific hypotheses for their functional follow-up.

Identifying and ranking potential driver genes of Alzheimer's disease using multiview evidence aggregation.

MOTIVATION: Late onset Alzheimer's disease is currently a disease with no known effective treatment options. To better understand disease, new multi-omic data-sets have recently been generated with the goal of identifying molecular causes of disease. However, most analytic studies using these datasets focus on uni-modal analysis of the data. Here, we propose a data driven approach to integrate multiple data types and analytic outcomes to aggregate evidences to support the hypothesis that a gene is a genetic driver of the disease. The main algorithmic contributions of our article are: (i) a general machine learning framework to learn the key characteristics of a few known driver genes from multiple feature sets and identifying other potential driver genes which have similar feature representations, and (ii) A flexible ranking scheme with the ability to integrate external validation in the form of Genome Wide Association Study summary statistics. While we currently focus on demonstrating the effectiveness of the approach using different analytic outcomes from RNA-Seq studies, this method is easily generalizable to other data modalities and analysis types. RESULTS: We demonstrate the utility of our machine learning algorithm on two benchmark multiview datasets by significantly outperforming the baseline approaches in predicting missing labels. We then use the algorithm to predict and rank potential drivers of Alzheimer's. We show that our ranked genes show a significant enrichment for single nucleotide polymorphisms associated with Alzheimer's and are enriched in pathways that have been previously associated with the disease. AVAILABILITY AND IMPLEMENTATION: Source code and link to all feature sets is available at https://github.com/Sage-Bionetworks/EvidenceAggregatedDriverRanking.

Unsupervised Analysis of Transcriptomics in Bacterial Sepsis Across Multiple Datasets Reveals Three Robust Clusters.

OBJECTIVES: To find and validate generalizable sepsis subtypes using data-driven clustering. DESIGN: We used advanced informatics techniques to pool data from 14 bacterial sepsis transcriptomic datasets from eight different countries (n = 700). SETTING: Retrospective analysis. SUBJECTS: Persons admitted to the hospital with bacterial sepsis. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: A unified clustering analysis across 14 discovery datasets revealed three subtypes, which, based on functional analysis, we termed "Inflammopathic, Adaptive, and Coagulopathic." We then validated these subtypes in nine independent datasets from five different countries (n = 600). In both discovery and validation data, the Adaptive subtype is associated with a lower clinical severity and lower mortality rate, and the Coagulopathic subtype is associated with higher mortality and clinical coagulopathy. Further, these clusters are statistically associated with clusters derived by others in independent single sepsis cohorts. CONCLUSIONS: The three sepsis subtypes may represent a unifying framework for understanding the molecular heterogeneity of the sepsis syndrome. Further study could potentially enable a precision medicine approach of matching novel immunomodulatory therapies with septic patients most likely to benefit.

Disentangling personalized treatment effects from "time-of-the-day" confounding in mobile health studies.

Ideally, a patient's response to medication can be monitored by measuring changes in performance of some activity. In observational studies, however, any detected association between treatment ("on-medication" vs "off-medication") and the outcome (performance in the activity) might be due to confounders. In particular, causal inferences at the personalized level are especially vulnerable to confounding effects that arise in a cyclic fashion. For quick acting medications, effects can be confounded by circadian rhythms and daily routines. Using the time-of-the-day as a surrogate for these confounders and the performance measurements as captured on a smartphone, we propose a personalized statistical approach to disentangle putative treatment and "time-of-the-day" effects, that leverages conditional independence relations spanned by causal graphical models involving the treatment, time-of-the-day, and outcome variables. Our approach is based on conditional independence tests implemented via standard and temporal linear regression models. Using synthetic data, we investigate when and how residual autocorrelation can affect the standard tests, and how time series modeling (namely, ARIMA and robust regression via HAC covariance matrix estimators) can remedy these issues. In particular, our simulations illustrate that when patients perform their activities in a paired fashion, positive autocorrelation can lead to conservative results for the standard regression approach (i.e., lead to deflated true positive detection), whereas negative autocorrelation can lead to anticonservative behavior (i.e., lead to inflated false positive detection). The adoption of time series methods, on the other hand, leads to well controlled type I error rates. We illustrate the application of our methodology with data from a Parkinson's disease mobile health study.

Remote smartphone monitoring of Parkinson's disease and individual response to therapy.

Remote health assessments that gather real-world data (RWD) outside clinic settings require a clear understanding of appropriate methods for data collection, quality assessment, analysis and interpretation. Here we examine the performance and limitations of smartphones in collecting RWD in the remote mPower observational study of Parkinson's disease (PD). Within the first 6 months of study commencement, 960 participants had enrolled and performed at least five self-administered active PD symptom assessments (speeded tapping, gait/balance, phonation or memory). Task performance, especially speeded tapping, was predictive of self-reported PD status (area under the receiver operating characteristic curve (AUC) = 0.8) and correlated with in-clinic evaluation of disease severity (r = 0.71; P < 1.8 × 10-6) when compared with motor Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS). Although remote assessment requires careful consideration for accurate interpretation of RWD, our results support the use of smartphones and wearables in objective and personalized disease assessments.

Crowdsourcing digital health measures to predict Parkinson's disease severity: the Parkinson's Disease Digital Biomarker DREAM Challenge.

Consumer wearables and sensors are a rich source of data about patients' daily disease and symptom burden, particularly in the case of movement disorders like Parkinson's disease (PD). However, interpreting these complex data into so-called digital biomarkers requires complicated analytical approaches, and validating these biomarkers requires sufficient data and unbiased evaluation methods. Here we describe the use of crowdsourcing to specifically evaluate and benchmark features derived from accelerometer and gyroscope data in two different datasets to predict the presence of PD and severity of three PD symptoms: tremor, dyskinesia, and bradykinesia. Forty teams from around the world submitted features, and achieved drastically improved predictive performance for PD status (best AUROC = 0.87), as well as tremor- (best AUPR = 0.75), dyskinesia- (best AUPR = 0.48) and bradykinesia-severity (best AUPR = 0.95).

Large eQTL meta-analysis reveals differing patterns between cerebral cortical and cerebellar brain regions.

The availability of high-quality RNA-sequencing and genotyping data of post-mortem brain collections from consortia such as CommonMind Consortium (CMC) and the Accelerating Medicines Partnership for Alzheimer's Disease (AMP-AD) Consortium enable the generation of a large-scale brain cis-eQTL meta-analysis. Here we generate cerebral cortical eQTL from 1433 samples available from four cohorts (identifying >4.1 million significant eQTL for >18,000 genes), as well as cerebellar eQTL from 261 samples (identifying 874,836 significant eQTL for >10,000 genes). We find substantially improved power in the meta-analysis over individual cohort analyses, particularly in comparison to the Genotype-Tissue Expression (GTEx) Project eQTL. Additionally, we observed differences in eQTL patterns between cerebral and cerebellar brain regions. We provide these brain eQTL as a resource for use by the research community. As a proof of principle for their utility, we apply a colocalization analysis to identify genes underlying the GWAS association peaks for schizophrenia and identify a potentially novel gene colocalization with lncRNA RP11-677M14.2 (posterior probability of colocalization 0.975).

Meta-Analysis of the Alzheimer's Disease Human Brain Transcriptome and Functional Dissection in Mouse Models.

We present a consensus atlas of the human brain transcriptome in Alzheimer's disease (AD), based on meta-analysis of differential gene expression in 2,114 postmortem samples. We discover 30 brain coexpression modules from seven regions as the major source of AD transcriptional perturbations. We next examine overlap with 251 brain differentially expressed gene sets from mouse models of AD and other neurodegenerative disorders. Human-mouse overlaps highlight responses to amyloid versus tau pathology and reveal age- and sex-dependent expression signatures for disease progression. Human coexpression modules enriched for neuronal and/or microglial genes broadly overlap with mouse models of AD, Huntington's disease, amyotrophic lateral sclerosis, and aging. Other human coexpression modules, including those implicated in proteostasis, are not activated in AD models but rather following other, unexpected genetic manipulations. Our results comprise a cross-species resource, highlighting transcriptional networks altered by human brain pathophysiology and identifying correspondences with mouse models for AD preclinical studies.

Dynamical analysis of cellular networks based on the Green's function matrix.

The complexity of cellular networks often limits human intuition in understanding functional regulations in a cell from static network diagrams. To this end, mathematical models of ordinary differential equations (ODEs) have commonly been used to simulate dynamical behavior of cellular networks, to which a quantitative model analysis can be applied in order to gain biological insights. In this paper, we introduce a dynamical analysis based on the use of Green's function matrix (GFM) as sensitivity coefficients with respect to initial concentrations. In contrast to the classical (parametric) sensitivity analysis, the GFM analysis gives a dynamical, molecule-by-molecule insight on how system behavior is accomplished and complementarily how (impulse) signal propagates through the network. The knowledge gained will have application from model reduction and validation to drug discovery research in identifying potential drug targets, studying drug efficacy and specificity, and optimizing drug dosing and timing. The efficacy of the method is demonstrated through applications to common network motifs and a Fas-induced programmed cell death model in Jurkat T cell line.

Detecting the impact of subject characteristics on machine learning-based diagnostic applications.

Collection of high-dimensional, longitudinal digital health data has the potential to support a wide-variety of research and clinical applications including diagnostics and longitudinal health tracking. Algorithms that process these data and inform digital diagnostics are typically developed using training and test sets generated from multiple repeated measures collected across a set of individuals. However, the inclusion of repeated measurements is not always appropriately taken into account in the analytical evaluations of predictive performance. The assignment of repeated measurements from each individual to both the training and the test sets ("record-wise" data split) is a common practice and can lead to massive underestimation of the prediction error due to the presence of "identity confounding." In essence, these models learn to identify subjects, in addition to diagnostic signal. Here, we present a method that can be used to effectively calculate the amount of identity confounding learned by classifiers developed using a record-wise data split. By applying this method to several real datasets, we demonstrate that identity confounding is a serious issue in digital health studies and that record-wise data splits for machine learning- based applications need to be avoided.

A community approach to mortality prediction in sepsis via gene expression analysis.

Improved risk stratification and prognosis prediction in sepsis is a critical unmet need. Clinical severity scores and available assays such as blood lactate reflect global illness severity with suboptimal performance, and do not specifically reveal the underlying dysregulation of sepsis. Here, we present prognostic models for 30-day mortality generated independently by three scientific groups by using 12 discovery cohorts containing transcriptomic data collected from primarily community-onset sepsis patients. Predictive performance is validated in five cohorts of community-onset sepsis patients in which the models show summary AUROCs ranging from 0.765-0.89. Similar performance is observed in four cohorts of hospital-acquired sepsis. Combining the new gene-expression-based prognostic models with prior clinical severity scores leads to significant improvement in prediction of 30-day mortality as measured via AUROC and net reclassification improvement index These models provide an opportunity to develop molecular bedside tests that may improve risk stratification and mortality prediction in patients with sepsis.

Molecular, phenotypic, and sample-associated data to describe pluripotent stem cell lines and derivatives.

The use of induced pluripotent stem cells (iPSC) derived from independent patients and sources holds considerable promise to improve the understanding of development and disease. However, optimized use of iPSC depends on our ability to develop methods to efficiently qualify cell lines and protocols, monitor genetic stability, and evaluate self-renewal and differentiation potential. To accomplish these goals, 57 stem cell lines from 10 laboratories were differentiated to 7 different states, resulting in 248 analyzed samples. Cell lines were differentiated and characterized at a central laboratory using standardized cell culture methodologies, protocols, and metadata descriptors. Stem cell and derived differentiated lines were characterized using RNA-seq, miRNA-seq, copy number arrays, DNA methylation arrays, flow cytometry, and molecular histology. All materials, including raw data, metadata, analysis and processing code, and methodological and provenance documentation are publicly available for re-use and interactive exploration at https://www.synapse.org/pcbc. The goal is to provide data that can improve our ability to robustly and reproducibly use human pluripotent stem cells to understand development and disease.

Gene expression elucidates functional impact of polygenic risk for schizophrenia.

Over 100 genetic loci harbor schizophrenia-associated variants, yet how these variants confer liability is uncertain. The CommonMind Consortium sequenced RNA from dorsolateral prefrontal cortex of people with schizophrenia (N = 258) and control subjects (N = 279), creating a resource of gene expression and its genetic regulation. Using this resource, ∼20% of schizophrenia loci have variants that could contribute to altered gene expression and liability. In five loci, only a single gene was involved: FURIN, TSNARE1, CNTN4, CLCN3 or SNAP91. Altering expression of FURIN, TSNARE1 or CNTN4 changed neurodevelopment in zebrafish; knockdown of FURIN in human neural progenitor cells yielded abnormal migration. Of 693 genes showing significant case-versus-control differential expression, their fold changes were ≤ 1.33, and an independent cohort yielded similar results. Gene co-expression implicates a network relevant for schizophrenia. Our findings show that schizophrenia is polygenic and highlight the utility of this resource for mechanistic interpretations of genetic liability for brain diseases.

TRIM32 modulates pluripotency entry and exit by directly regulating Oct4 stability.

Induced pluripotent stem cells (iPSCs) have revolutionized the world of regenerative medicine; nevertheless, the exact molecular mechanisms underlying their generation and differentiation remain elusive. Here, we investigated the role of the cell fate determinant TRIM32 in modulating such processes. TRIM32 is essential for the induction of neuronal differentiation of neural stem cells by poly-ubiquitinating cMyc to target it for degradation resulting in inhibition of cell proliferation. To elucidate the role of TRIM32 in regulating somatic cell reprogramming we analysed the capacity of TRIM32-knock-out mouse embryonic fibroblasts (MEFs) in generating iPSC colonies. TRIM32 knock-out MEFs produced a higher number of iPSC colonies indicating a role for TRIM32 in inhibiting this cellular transition. Further characterization of the generated iPSCs indicated that the TRIM32 knock-out iPSCs show perturbed differentiation kinetics. Additionally, mathematical modelling of global gene expression data revealed that during differentiation an Oct4 centred network in the wild-type cells is replaced by an E2F1 centred network in the TRIM32 deficient cells. We show here that this might be caused by a TRIM32-dependent downregulation of Oct4. In summary, the data presented here reveal that TRIM32 directly regulates at least two of the four Yamanaka Factors (cMyc and Oct4), to modulate cell fate transitions.

Integrating pathways of Parkinson's disease in a molecular interaction map.

Parkinson's disease (PD) is a major neurodegenerative chronic disease, most likely caused by a complex interplay of genetic and environmental factors. Information on various aspects of PD pathogenesis is rapidly increasing and needs to be efficiently organized, so that the resulting data is available for exploration and analysis. Here we introduce a computationally tractable, comprehensive molecular interaction map of PD. This map integrates pathways implicated in PD pathogenesis such as synaptic and mitochondrial dysfunction, impaired protein degradation, alpha-synuclein pathobiology and neuroinflammation. We also present bioinformatics tools for the analysis, enrichment and annotation of the map, allowing the research community to open new avenues in PD research. The PD map is accessible at http://minerva.uni.lu/pd_map .

Detecting cellular reprogramming determinants by differential stability analysis of gene regulatory networks.

BACKGROUND: Cellular differentiation and reprogramming are processes that are carefully orchestrated by the activation and repression of specific sets of genes. An increasing amount of experimental results show that despite the large number of genes participating in transcriptional programs of cellular phenotypes, only few key genes, which are coined here as reprogramming determinants, are required to be directly perturbed in order to induce cellular reprogramming. However, identification of reprogramming determinants still remains a combinatorial problem, and the state-of-art methods addressing this issue rests on exhaustive experimentation or prior knowledge to narrow down the list of candidates. RESULTS: Here we present a computational method, without any preliminary selection of candidate genes, to identify reduced subsets of genes, which when perturbed can induce transitions between cellular phenotypes. The method relies on the expression profiles of two stable cellular phenotypes along with a topological analysis stability elements in the gene regulatory network that are necessary to cause this multi-stability. Since stable cellular phenotypes can be considered as attractors of gene regulatory networks, cell fate and cellular reprogramming involves transition between these attractors, and therefore current method searches for combinations of genes that are able to destabilize a specific initial attractor and stabilize the final one in response to the appropriate perturbations. CONCLUSIONS: The method presented here represents a useful framework to assist researchers in the field of cellular reprogramming to design experimental strategies with potential applications in the regenerative medicine and disease modelling.

Understanding dynamics using sensitivity analysis: caveat and solution.

BACKGROUND: Parametric sensitivity analysis (PSA) has become one of the most commonly used tools in computational systems biology, in which the sensitivity coefficients are used to study the parametric dependence of biological models. As many of these models describe dynamical behaviour of biological systems, the PSA has subsequently been used to elucidate important cellular processes that regulate this dynamics. However, in this paper, we show that the PSA coefficients are not suitable in inferring the mechanisms by which dynamical behaviour arises and in fact it can even lead to incorrect conclusions. RESULTS: A careful interpretation of parametric perturbations used in the PSA is presented here to explain the issue of using this analysis in inferring dynamics. In short, the PSA coefficients quantify the integrated change in the system behaviour due to persistent parametric perturbations, and thus the dynamical information of when a parameter perturbation matters is lost. To get around this issue, we present a new sensitivity analysis based on impulse perturbations on system parameters, which is named impulse parametric sensitivity analysis (iPSA). The inability of PSA and the efficacy of iPSA in revealing mechanistic information of a dynamical system are illustrated using two examples involving switch activation. CONCLUSIONS: The interpretation of the PSA coefficients of dynamical systems should take into account the persistent nature of parametric perturbations involved in the derivation of this analysis. The application of PSA to identify the controlling mechanism of dynamical behaviour can be misleading. By using impulse perturbations, introduced at different times, the iPSA provides the necessary information to understand how dynamics is achieved, i.e. which parameters are essential and when they become important.