Biology subjects

Zaitlen, N.

Publications and source records attributed to Zaitlen, N..

At least 37 records · Page 2Linked to original sources

bioRxiv2021

Seeing beyond the target: Leveraging off-target reads in targeted clinical tumor sequencing to identify prognostic biomarkers

Clinical tumor sequencing is rapidly becoming a standard component of clinical care, providing essential information for selecting amongst treatment options and providing prognostic value. Here we develop a robust and scalable software platform (SBT: Seeing Beyond the Target) that mines discarded components of clinical sequences to produce estimates of a rich set of omics features including rDNA and mtDNA copy number, microbial species abundance, and T and B cell receptor sequences. We validate the accuracy of SBT via comparison to multimodal data from the TCGA and apply SBT to a tumor panel cohort of 2,920 lung adenocarcinomas to identify associations of clinical value. We replicated known associations of somatic events in TP53 with changes in rDNA (p=0.012); as well as diversity of BCR and TCR repertoires with the biopsy site (p=2.5x10-6, p<10-20). We observed striking differences in EGFR mutant lung cancers versus wild-type, including higher rDNA copy number and lower immune repertoire diversity. Integrating clinical outcomes, we identified significant prognostic associations with overall survival, including SBT estimates of 5S rDNA (p=1.9x10-4, hazard ratio = 1.22) and TCR diversity (p=2.7x10-3, hazard ratio=1.77). Both novel survival associations replicated in 1,302 breast carcinoma and 1,651 colorectal cancer tumors. We anticipate that feature estimates derived by SBT will yield novel biomarker hypotheses and open research opportunities in existing and emerging clinical tumor sequencing cohorts.

genomics↗

bioRxiv2020

Leveraging pleiotropy in genome-wide association studies in multiple traits with per trait interpretations

We introduce pleiotropic association test (PAT) for joint analysis of multiple traits using GWAS summary statistics. The method utilizes the decomposition of phenotypic covariation into genetic and environmental components to create a likelihood ratio test statistic for each genetic variant. Though PAT does not directly interpret which trait(s) drive the association, a per trait interpretation of the omnibus p-value is provided through an extension to the meta-analysis framework, m-values. In simulations, we show PAT controls the false positive rate, increases statistical power, and is robust to model misspecifications of genetic effect. Additionally, simulations comparing PAT to two multi-trait methods, HIPO and MTAG show PAT having a 43.0% increase in the number of omnibus associations over the other methods. When these associations are interpreted on a per trait level using m-values, PAT has 52.2% more per trait interpretations with a 0.57% false positive assignment rate. When analyzing four traits from the UK Biobank, PAT identifies 22,095 novel associated variants. Through the m-values interpretation framework, the number of total per trait associations for two traits are almost tripled and are nearly doubled for another trait relative to the original single trait GWAS.

bioinformatics↗

bioRxiv2020

Coordinated Interaction: A model and test for globally signed epistasis in complex traits

Interactions between genetic variants - epistasis - is pervasive in model systems and can profoundly impact evolutionary adaption, population disease dynamics, genetic mapping, and precision medicine efforts. In this work we develop a model for structured polygenic epistasis, called Coordinated Interaction (CI), and prove that several recent theories of genetic architecture fall under the formal umbrella of CI. Unlike standard polygenic epistasis models that assume interaction and main effects are independent, in the CI model, sets of SNPs broadly interact positively or negatively, on balance skewing the penetrance of main genetic effects. To test for the existence of CI we propose the even-odd (EO) test and prove it is calibrated in a range of realistic biological models. Applying the EO test in the UK Biobank, we find evidence of CI in 14 of 26 traits spanning disease, anthropometric, and blood categories. Finally, we extend the EO test to tissue-specific enrichment and identify several plausible tissue-trait pairs. Overall, CI is a new dimension of genetic architecture that can capture structured, systemic interactions in complex human traits.

genetics↗

bioRxiv2020

MPRAudit Quantifies the Fraction of Variance Described by Unknown Features in Massively Parallel Reporter Assays

Transformative advances in molecular technologies, such as massively parallel reporter assays (MPRAs) and CRISPR screens, can efficiently characterize the effects of genetic and genomic variation on cellular phenotypes. Analysis approaches to date have focused on identifying individual genomic regions or genetic variants that perturb a phenotype of interest. In this work, we develop a wholistic framework (MPRAudit) to determine the global contribution of sequence to phenotypic variation across subsets of the entire experiment, opening the door to myriad novel analyses. For example, MPRAudit can reliably estimate the upper limit of predictive performance, the fraction of variation attributed to specific biological categories, and the total contribution of experimental noise. We demonstrate through simulation and application to several types of real MPRA data sets how MPRAudit can lead to an improved understanding of experimental quality, molecular biology, and guide future research. Applying MPRAudit to real MPRA data, we observe that sequence variation is the primary driver of outcome variability, but that known biological categories explain only a fraction of this variance. We conclude that our understanding of how sequence variation impacts phenotype, even at the level of MPRAs, remains open to further scientific discovery.

bioinformatics↗

bioRxiv2020

Massively Parallel Analysis of Human 3' UTRs Reveals that AU-Rich Element Length and Registration Predict mRNA Destabilization

AU-rich elements (AREs) are 3' UTR cis-regulatory elements that regulate the stability of mRNAs. Consensus ARE motifs have been determined, but little is known about how differences in 3' UTR sequences that conform to these motifs affect their function. Here we use functional annotation of sequences from 3' UTRs (fast-UTR), a massively parallel reporter assay (MPRA), to investigate the effects of 41,288 3' UTR sequence fragments from 4,653 transcripts on gene expression and mRNA stability. The library included 9,142 AREs, and incorporated a set of fragments bearing mutations in each ARE. Our analyses demonstrate that the length of an ARE and its registration (the first and last nucleotides of the repeating ARE motif) have significant effects on gene expression and stability. Based on this finding, we propose improved ARE classification and concomitant methods to categorize and predict the effect of AREs on gene expression and stability. Our new approach explains 64{+/-}13% of the contribution of AREs to the stability of human 3' UTRs in Jurkat cells and predicts ARE activity in an unrelated cell type. Finally, to investigate the advantages of our general experimental design for annotating 3' UTR elements we examine other motifs including constitutive decay elements (CDEs), where we show that the length of the CDE stem-loop has a significant impact on steady-state expression and mRNA stability. We conclude that fast-UTR, in conjunction with our analytical approach, can produce improved yet simple sequence-based rules for predicting the activity of human 3' UTRs containing functional motifs.

genomics↗

bioRxiv2020

Estimating the rate of cell type degeneration from epigenetic sequencing of cell-free DNA

Circulating cell-free DNA (cfDNA) in the bloodstream originates from dying cells and is a promising non-invasive biomarker for cell death. Here, we develop a method to accurately estimate the relative abundances of cell types contributing to cfDNA. We leverage the distinct DNA methylation profile of each cell type throughout the body. Decomposing the cfDNA mixture is difficult, as fragments from relevant cell types may only be present in a small amount. We propose an algorithm, CelFiE, that estimates cell type proportion from both whole genome cfDNA input and reference data. CelFiE accommodates low coverage data, does not rely on CpG site curation, and estimates contributions from multiple unknown cell types that are not available in reference data. In simulations we show that CelFiE can accurately estimate known and unknown cell type of origin of cfDNA mixtures in low coverage and noisy data. Simulations also demonstrate that we can effectively estimate cfDNA originating from rare cell types composing less than 0.01% of the total cfDNA. To validate CelFiE, we use a positive control: cfDNA extracted from pregnant and non-pregnant women. CelFiE estimates a large placenta component specifically in pregnant women (p = 9.1 x 10-5). Finally, we use CelFiE to decompose cfDNA from ALS patients and age matched controls. We find increased cfDNA concentrations in ALS patients (p = 3.0 x 10-3). Specifically, CelFiE estimates increased skeletal muscle component in the cfDNA of ALS patients (p = 2.6 x 10-3), which is consistent with muscle impairment characterizing ALS. Quantification of skeletal muscle death in ALS is novel, and overall suggests that CelFiE may be a useful tool for biomarker discovery and monitoring of disease progression.

genomics↗

bioRxiv2019

Whole-genome and RNA sequencing reveal variation and transcriptomic coordination in the developing human prefrontal cortex

Variation in gene expression underlies neurotypical development, while genomic variants contribute to neuropsychiatric disorders. BrainVar is a unique resource of paired whole-genome sequencing and bulk-tissue RNA-sequencing from the human dorsolateral prefrontal cortex of 176 neurotypical individuals across prenatal and postnatal development, providing the opportunity to assay genomic and transcriptomic variation in tandem. Leveraging this resource, we identified rare premature stop codons with commensurate reduced and allele-specific expression of corresponding genes, and common variants that alter gene expression (expression quantitative trait loci, eQTLs). Categorizing eQTLs by prenatal and postnatal effect, genes affected by temporally-specific eQTLs, compared to constitutive eQTLs, are enriched for haploinsufficiency, protein-protein interactions, and neuropsychiatric disorder risk loci. Expression levels of over 12,000 genes rise or fall in a concerted late-fetal transition, with the transitional genes enriched for cell type specific genes and neuropsychiatric disorder loci, underscoring the importance of cataloguing developmental trajectories in understanding cortical physiology and pathology.\n\nHighlightsO_LIWhole-genome and RNA-sequencing across human prefrontal cortex development in BrainVar\nC_LIO_LIGene-specific developmental trajectories characterize the late-fetal transition\nC_LIO_LIIdentification of constitutive, prenatal-specific, postnatal-specific, and rare eQTLs\nC_LIO_LIIntegrated analysis reveals genetic and developmental influences on CNS traits and disorders\nC_LI

neuroscience↗

bioRxiv2019

On the cross-population portability of gene expression prediction models

The genetic control of gene expression is a core component of human physiology. For the past several years, transcriptome-wide association studies have leveraged large datasets of linked genotype and RNA sequencing information to create a powerful gene-based test of association that has been used in dozens of studies. While numerous discoveries have been made, the populations in the training data are overwhelmingly of European descent, and little is known about the generalizability of these models to other populations. Here, we test for cross-population generalizability of gene expression prediction models using a dataset of African American individuals with RNA-Seq data in whole blood. We find that the default models trained in large datasets such as GTEx and DGN fare poorly in African Americans, with a notable reduction in prediction accuracy when compared to European Americans. We replicate these limitations in cross-population generalizability using the five populations in the GEUVADIS dataset. Via realistic simulations of both populations and gene expression, we show that accurate cross-population generalizability of transcriptome prediction only arises when eQTL architecture is substantially shared across populations. In contrast, models with non-identical eQTLs showed patterns similar to real-world data. Therefore, generating RNA-Seq data in diverse populations is a critical step towards multi-ethnic utility of gene expression prediction. Author summaryAdvances in RNA sequencing technology have reduced the cost of measuring gene expression at a genome-wide level. However, sequencing enough human RNA samples for adequately-powered disease association studies remains prohibitively costly. To this end, modern transcriptome-wide association analysis tools leverage existing paired genotype-expression datasets by creating models to predict gene expression using genotypes. These predictive models enable researchers to perform cost-effective association tests with gene expression in independently genotyped samples. However, most of these models use European reference data, and the extent to which gene expression prediction models work across populations is not fully resolved. We observe that these models predict gene expression worse than expected in a dataset of African-Americans when derived from European-descent individuals. Using simulations, we show that gene expression predictive model performance depends on both the amount of shared genotype predictors as well as the genetic relatedness between populations. Our findings suggest a need to carefully select reference populations for prediction and point to a pressing need for more genetically diverse genotype-expression datasets.

genomics↗

bioRxiv2019

The Effects of Mutations are Modified by Genetic Background in Mice

BackgroundThe majority of quantitative genetic models used to map complex traits assume that alleles have similar effects across all individuals. Significant evidence suggests, however, that epistatic interactions modulate the impact of many alleles. Nevertheless, identifying epistatic interactions remains computationally and statistically challenging. In this work, we address some of these challenges by developing a statistical test for polygenic epistasis that determines whether the effect of an allele is altered by the global genetic ancestry proportion from distinct progenitors. ResultsWe applied our method to data from mice and yeast. For the mice, we observed 49 significant genotype-by-ancestry interaction associations across 14 phenotypes as well as over 1,400 Bonferroni-corrected genotype-by-ancestry interaction associations for mouse gene expression data. For the yeast, we observed 92 significant genotype-by-ancestry interactions across 38 phenotypes. Given this evidence of epistasis, we test for and observe evidence of rapid selection pressure on ancestry specific polymorphisms within one of the cohorts, consistent with epistatic selection. ConclusionsUnlike our prior work in human populations, we observe widespread evidence of ancestry-modified SNP effects, perhaps reflecting the greater divergence present in crosses using mice and yeast. Author SummaryMany statistical tests which link genetic markers in the genome to differences in traits rely on the assumption that the same polymorphism will have identical effects in different individuals. However, there is substantial evidence indicating that this is not the case. Epistasis is the phenomenon in which multiple polymorphisms interact with one another to amplify or negate each others effects on a trait. We hypothesized that individual SNP effects could be changed in a polygenic manner, such that the proportion of as genetic ancestry, rather than specific markers, might be used to capture epistatic interactions. Motivated by this possibility, we develop a new statistical test that allowed us to examine the genome to identify polymorphisms which have different effects depending on the ancestral makeup of each individual. We use our test in two different populations of inbred mice and a yeast panel and demonstrate that these sorts of variable effect polymorphisms exist in 14 different physical traits in mice and 38 phenotypes in yeast as well as in murine gene expression. We use the term "polygenic epistasis" to distinguish these interactions from the more conventional two- or multi-locus interactions.

genomics↗

bioRxiv2019

Efficient estimation and applications of cross-validated genetic predictions

Large-scale cohorts with combined genetic and phenotypic data, coupled with methodological advances, have produced increasingly accurate genetic predictors of complex human phenotypes called polygenic risk scores (PRS). In addition to the potential translational impacts of identifying at-risk individuals, PRS are being utilized for a growing list of scientific applications including causal inference, identifying pleiotropy and genetic correlation, and powerful gene-based and mixed model association tests. Existing PRS approaches rely on external large-scale genetic cohorts that have also measured the phenotype of interest. They further require matching on ancestry and genotyping platform or imputation quality. In this work we present a novel reference-free method to produce PRS that does not rely on an external cohort. We show that naive implementations of reference-free PRS either result in substantial over-fitting or prohibitive increases in computational time. We show that our algorithm avoids both of these issues, and can produce informative in-sample PRS over any existing cohort without over-fitting. We then demonstrate several novel applications of reference-free PRS including detection of pleiotropy across 246 metabolic traits and efficient mixed-model association testing.

genetics↗

bioRxiv2018

A comprehensive study of metabolite genetics reveals strong pleiotropy and heterogeneity across time and context

Genetic studies of metabolites have identified thousands of variants many of which are associated with downstream metabolic and obesogenic disorders. However, these studies have relied on univariate analyses, reducing power and limiting context specific understanding. Here we aim to provide an integrated perspective of the genetic basis of metabolites by leveraging the Finnish Metabolic Syndrome In Men (METSIM) cohort, a unique genetic resource which contains metabolic measurements across distinct timepoints as well as detailed information on statin usage. We increase effective sample size by an average of two-fold by applying the Covariates for Multi-phenotype Studies (CMS) approach, identifying 588 significant SNP-metabolite associations, including 248 novel associations. We further show that many of these SNPs are master metabolic regulators, balancing the relative proportion of dozens of metabolite levels. We then identify the first associations to changes in metabolic levels across time as well as evidence of genetic interaction with statin use. Finally, we show an overall decrease in genetic control of metabolic processes with age.

genetics↗

bioRxiv2018

A global map of RNA binding protein occupancy guides functional dissection of post-transcriptional regulation of the T cell transcriptome

RNA binding proteins (RBPs) mediate constitutive RNA metabolism and gene specific regulatory interactions. To identify RNA cis-regulatory elements, we developed GCLiPP, a biochemical technique for detecting RBP occupancy transcriptome-wide. GCLiPP sequence tags corresponded with known RBP binding sites, specifically correlating to abundant cytosolic RBPs. To demonstrate the utility of our occupancy profiles, we performed functional dissection of 3' UTRs with CRISPR/Cas9 genome editing. Two RBP occupied sites in the CD69 3' UTR destabilized the transcript of this key regulator of lymphocyte tissue egress. Comparing human Jurkat T cells and mouse primary T cells uncovered hundreds of biochemically shared peaks of GCLiPP signal across homologous regions of human and mouse 3' UTRs, including a cis-regulatory element that governs the stability of the mRNA that encodes the proto-oncogene PIM3 in both species. Our GCLiPP datasets provide a rich resource for investigation of post-transcriptional regulation in the immune system.

genomics↗

bioRxiv2018

Reverse GWAS: Using Genetics to Identify and Model Phenotypic Subtypes

Recent and classical work has revealed biologically and medically significant subtypes in complex diseases and traits. However, relevant subtypes are often unknown, unmeasured, or actively debated, making automatic statistical approaches to subtype definition particularly valuable. We propose reverse GWAS (RGWAS) to identify and validate subtypes using genetics and multiple traits: while GWAS seeks the genetic basis of a given trait, RGWAS seeks to define trait subtypes with distinct genetic bases. Unlike existing approaches relying on off-the-shelf clustering methods, RGWAS uses a bespoke decomposition, MFMR, to model covariates, binary traits, and population structure. We use extensive simulations to show these features can be crucial for power and calibration. We validate RGWAS in practice by recovering known stress subtypes in major depressive disorder. We then show the utility of RGWAS by identifying three novel subtypes of metabolic traits. We biologically validate these metabolic subtypes with SNP-level tests and a novel polygenic test: the former recover known metabolic GxE SNPs; the latter suggests genetic heterogeneity may explain substantial missing heritability. Crucially, statins, which are widely prescribed and theorized to increase diabetes risk, have opposing effects on blood glucose across metabolic subtypes, suggesting potential have potential translational value.\n\nAuthor summaryComplex diseases depend on interactions between many known and unknown genetic and environmental factors. However, most studies aggregate these strata and test for associations on average across samples, though biological factors and medical interventions can have dramatically different effects on different people. Further, more-sophisticated models are often infeasible because relevant sources of heterogeneity are not generally known a priori. We introduce Reverse GWAS to simultaneously split samples into homogeneoues subtypes and to learn differences in genetic or treatment effects between subtypes. Unlike existing approaches to computational subtype identification using high-dimensional trait data, RGWAS accounts for covariates, binary disease traits and, especially, population structure; these features are each invaluable in extensive simulations. We validate RGWAS by recovering known genetic subtypes of major depression. We demonstrate RGWAS is practically useful in a metabolic study, finding three novel subtypes with both SNP- and polygenic-level heterogeneity. Importantly, RGWAS can uncover differential treatment response: for example, we show that statin, a common drug and potential type 2 diabetes risk factor, may have opposing subtype-specific effects on blood glucose.

genetics↗

bioRxiv2018

Existence and implications of population variance structure

Identifying the genetic and environmental factors underlying phenotypic differences between populations is fundamental to multiple research communities. To date, studies have focused on the relationship between population and phenotypic mean. Here we consider the relationship between population and phenotypic variance, i.e., \"population variance structure.\" In addition to gene-gene and gene-environment interaction, we show that population variance structure is a direct consequence of natural selection. We develop the ancestry double generalized linear model (ADGLM), a statistical framework to jointly model population mean and variance effects. We apply ADGLM to several deeply phenotyped datasets and observe ancestry-variance associations with 12 of 44 tested traits in ~113K British individuals and 3 of 14 tested traits in ~3K Mexican, Puerto Rican, and African-American individuals. We show through extensive simulations that population variance structure can both bias and reduce the power of genetic association studies, even when principal components or linear mixed models are used. ADGLM corrects this bias and improves power relative to previous methods in both simulated and real datasets. Additionally, ADGLM identifies 17 novel genotype-variance associations across six phenotypes.

genetics↗

bioRxiv2018

GxEMM: Extending linear mixed models to general gene-environment interactions

Gene-environment interaction (GxE) is a well-known source of non-additive inheritance. GxE can be important in applications ranging from basic functional genomics to precision medical treatment. Further, GxE effects elude inherently-linear LMMs and may explain missing heritability. We propose a simple, unifying mixed model for polygenic interactions (GxEMM) to capture the aggregate effect of small GxE effects spread across the genome. GxEMM extends existing LMMs for GxE in two important ways. First, it extends to arbitrary environmental variables, not just categorical groups. Second, GxEMM can estimate and test for environment-specific heritability. In simulations where the assumptions of existing methods do not hold, we show that GxEMM improves estimates of ordinary and GxE heritability and increases power to test for polygenic GxE. We then use GxEMM to prove that the heritability of major depression (MD) is reduced by stress, which we previously conjectured but could not prove with prior methods, and that a tail of polygenic GxE effects remains unexplained by MD GWAS.

genetics↗

bioRxiv2018

GBAT: a gene-based association method for robust trans-gene regulation detection

Identification of trans-eQTLs has been limited by a heavy multiple testing burden, read-mapping biases, and hidden confounders. To address these issues, we developed GBAT, a powerful gene-based method that allows robust detection of trans gene regulation. Using simulated and real data, we show that GBAT drastically increases detection of trans-gene regulation over standard trans-eQTL analyses.

genetics↗

bioRxiv2018

Genetic and environmental perturbations lead to regulatory decoherence

Correlation among traits is a fundamental feature of biological systems. From morphological characters, to transcriptional or metabolic networks, the correlations we routinely observe between traits reflect a shared regulation that remains poorly understood and difficult to study. To address this problem, we developed a new and flexible approach that allows us to identify factors associated with variation in correlation between individuals. Here, we use data from three large human cohorts to study the effects of genetic variation and environmental perturbation on correlations among mRNA transcripts and among NMR metabolites. We first show that environmental exposures (namely, infection and disease) lead to a systematic loss of correlation, which we define as decoherence. Using longitudinal data, we show that decoherent metabolites are better predictors of whether someone will develop metabolic syndrome than metabolites commonly used as biomarkers of this disease. Finally, we show that correlation itself is a trait under genetic control: specifically, we mapped and replicated hundreds of correlation QTLs, which often involve transcription factors or their known target genes. Together, this work furthers our understanding of how and why coordinated biological processes break down, and highlights the role of decoherence in disease emergence.

genomics↗

bioRxiv2017

Tracing cellular heterogeneity in pooled genetic screens via multi-level barcoding

While pooled loss- and gain-of-function screening approaches have become increasingly popular to systematically investigate mammalian gene function, they have thus far ignored the fact that cell populations are heterogeneous. Here we introduce multi-level barcoded sgRNA libraries to (i) monitor differences in the behavior of multiplexed clonal cell lines, (ii) trace sub-clonal lineages of cells expressing the same sgRNA, (iii) derive in-sample screen replicates and (iv) reduce the number of cells and sequencing read counts required to reach statistical significance. Using our approach, we illustrate how clonal heterogeneity impairs the results of pooled genetic screens and demonstrate the ability of multi-level barcoding to resolve cellular heterogeneity related issues.

genomics↗