Biology subjects

Mostafavi, S.

Publications and source records attributed to Mostafavi, S..

At least 37 records · Page 2Linked to original sources

bioRxiv2021

Mosaic loss of chromosome Y in aged human microglia.

Mosaic loss of chromosome Y (LOY) is a particularly common acquired structural mutation in the leukocytes of aging men and it has been shown to correlate with several age-related diseases including Alzheimers disease (AD). To derive the molecular basis of LOY in brain cells, we create an integrated resource by aggregating data from 21 single-cell and single-nuclei RNA brain studies, yielding 763,410 cells to investigate the presence and cell-type specific burden of LOY. We created robust quantification metrics for assessing LOY, which were validated using a multi-modal dataset. Using this new resource and LOY-quantification approach, we found that LOY frequencies differed widely between CNS cell-types and individual donors. Among five common neural cell types, microglia were most affected by LOY (7.79%, n=41,949), while LOY in neurons was rare (0.48%, n=220,010). Differential gene expression analysis in microglia found 188 autosomal genes, 6 X-linked genes, and 11 pseudoautosomal genes, pointing to broad dysregulation in lipoprotein metabolism, inflammatory response, and antigen processing that coincides with loss of Y. To our knowledge, we provide the first evidence of LOY in the microglia, and highlight its potential roles in aging and the pathogenesis of neurodegenerative disorders such as AD.

genomics↗

bioRxiv2021

In silico discovery of small molecules for efficient stem cell differentiation into definitive endoderm

Improving methods for human embryonic stem cell differentiation represents a challenge in modern regenerative medicine research. Using drug repurposing approaches, we discover small molecules that regulate the formation of definitive endoderm. Among them are inhibitors of known processes involved in endoderm differentiation (mTOR, PI3K, and JNK pathways) and a new compound, with an unknown mechanism of action, capable of inducing endoderm formation in the absence of growth factors in the media. Optimization of the classical protocol by including this compound achieves the same differentiation efficiency with a 90% cost reduction. The gene expression profile induced by the compound suggests that it is an inhibitor of the MYC pathway. The proposed in silico procedure for candidate molecule selection has broad potential for improving stem cell differentiation protocols.

bioinformatics↗

bioRxiv2021

Distinct mitochondrial remodeling during early cardiomyocyte development in a human-based stem cell model

Given the considerable interest in using stem cells for modelling and treating disease, it is essential to understand what regulates self-renewal and differentiation. Remodeling of mitochondria and metabolism, with the shift from glycolysis to oxidative phosphorylation (OXPHOS), play a fundamental role in maintaining pluripotency and stem cell fate. It has been suggested that metabolic switch from glycolysis to OXPHOS is germ-layer specific as during early ectoderm commitment, glycolysis remains active while during the transition to mesoderm and endoderm lineages, it is downregulated. How mitochondria adapt during these metabolic changes and whether mitochondria remodeling is tissue specific remains unclear. Here we address the question of mitochondrial adaption by examining the differentiation of human pluripotent stem cells to cardiac progenitors and further to functional cardiomyocytes. Contrary to recent findings in neuronal differentiation, we found that mitochondrial content decreases continuously during mesoderm differentiation, despite clear mitochondrial remodeling giving increased mitochondrial activity and higher levels of ATP-linked respiration. Thus, our work both highlights similarities in mitochondrial remodeling during the transition from pluripotent to multipotent state in ectodermal and mesodermal lineages, while at the same time demonstrating cell-lineage-specific adaptions upon further differentiation. Our results improve understanding of how mitochondria remodeling and the metabolism interact during differentiation and show that it is erroneous to assume that increased OXPHOS activity during differentiation requires a simultaneous expansion of mitochondrial content. Summary statementWe found that mitochondrial content decreases continuously during mesoderm differentiation, despite clear mitochondrial remodeling giving increased mitochondrial activity and higher levels of ATP-linked respiration during mesoderm differentiation.

cell biology↗

bioRxiv2020

Biologically-relevant transfer learning improves transcription factor binding prediction

BackgroundDeep learning has proven to be a powerful technique for transcription factor (TF) binding prediction, but requires large training datasets. Transfer learning can reduce the amount of data required for deep learning, while improving overall model performance, compared to training a separate model for each new task. ResultsWe assess a transfer learning strategy for TF binding prediction consisting of a pre-training step, wherein we train a multi-task model with multiple TFs, and a fine-tuning step, wherein we initialize single-task models for individual TFs with the weights learned by the multi-task model, after which the single-task models are trained at a lower learning rate. We corroborate that transfer learning improves model performance, especially if in the pre-training step the multi-task model is trained with biologically-relevant TFs. We show the effectiveness of transfer learning for TFs with [~]500 ChIP-seq peak regions. Using model interpretation techniques, we demonstrate that the features learned in the pre-training step are refined in the fine-tuning step to resemble the binding motif of the target TF (i.e. the recipient of transfer learning in the fine-tuning step). Moreover, pre-training with biologically-relevant TFs allows single-task models in the fine-tuning step to learn features other than the motif of the target TF. ConclusionsOur results confirm that transfer learning is a powerful technique for TF binding prediction.

bioinformatics↗

bioRxiv2020

Unified AI framework to uncover deep interrelationships between gene expression and Alzheimer's disease neuropathologies

Deep neural networks offer a promising approach for capturing complex, non-linear relationships among variables. Because they require immense sample sizes, their potential has yet to be fully tapped for understanding complex relationships between gene expression and human phenotypes. Encouragingly, a growing number of diseases are being studied through consortium efforts. Here we introduce a new analysis framework, namely MD-AD (Multi-task Deep learning for Alzheimers Disease neuropathology), which leverages an unexpected synergy between deep neural networks and multi-cohort settings. In these settings, true joint analysis can be stymied using conventional statistical methods, which (1) require "harmonized" phenotypes (i.e., measured in a highly consistent manner) and (2) tend to capture cohort-level variations, obscuring the subtler true disease signals. Instead, MD-AD incorporates multiple related phenotypes sparsely measured across cohorts, and learns complex, non-linear interactions between genes and phenotypes not discovered using conventional expression data analysis methods (e.g., component analysis and module detection), enabling the model to capture subtler signals than cohort-level variations. Applied to the largest available collection of brain samples (N=1,758), we demonstrate that MD-AD learns a truly generalizable relationship between gene expression program and AD-related neuropathology. The learned program generalizes in several important ways, including recapitulation of the disease progress in animal models and across tissue types, and we show that such generalizability is not achieved by previous statistical paradigms. Its ability to identify genes with high non-linear relevance to neuropathology enabled us to identify a sex-specific relationship between neuropathology and immune response across microglia, providing a nuanced context for association between inflammatory genes and AD.

systems biology↗

bioRxiv2020

Decoding differential gene expression

Identifying the molecular mechanisms that control differential gene expression (DE) is a major goal of basic and disease biology. Combining the strengths of systems biology and deep learning in a model called DEcode, we are able to predict DE more accurately than traditional sequence-based methods, which do not utilize systems biology data. To determine the biological origins of this accuracy, we identify the most predictive regulators and types of regulatory interactions in DEcode, contrasting their roles across many human tissues. Diverse systems biology, ontological and disease-related assessments all point to the predominant influence of post-translational RNA-binding factors on DE. Through the combinatorial gene regulation that is captured in DEcode, it is even possible to predict relatively subtle person-to-person variation in gene expression. We demonstrate the broad applicability of these clinically-relevant predictions by predicting drivers of aging throughout the human lifespan, gene coexpression relationships on a genome-wide scale, and frequent DE in diverse conditions. Researchers can freely access DEcode to utilize genomic big data in identifying influential molecular mechanisms for any human expression data - www.differentialexpression.org.

systems biology↗

bioRxiv2019

Learning immune cell differentiation

The mammalian genome contains several million cis-regulatory elements, whose differential activity marked by open chromatin determines organogenesis and differentiation. This activity is itself embedded in the DNA sequence, decoded by sequence-specific transcription factors. Leveraging a granular ATAC-seq atlas of chromatin activity across 81 immune cell-types we show that a convolutional neural network ("AI-TAC") can learn to infer cell-type-specific chromatin activity solely from the DNA sequence. AI-TAC does so by rediscovering, with astonishing precision, binding motifs for known regulators, and some unknown ones, mapping them with high concordance to positions validated by ChIP-seq data. AI-TAC also uncovers combinatorial influences, establishing a hierarchy of transcription factors (TFs) and their interactions involved in immunocyte specification, with intriguingly different strategies between lineages. Mouse-trained AI-TAC can parse human DNA, revealing a strikingly similar ranking of influential TFs. Thus, Deep Learning can reveal the regulatory syntax that drives the full differentiative complexity of the immune system.

genomics↗

bioRxiv2019

Cascading Epigenomic Model for GWAS

The majority of genetic variants detected in genome wide association studies (GWAS) exert their effects on phenotypes through gene regulation. Motivated by this observation, we propose a multi-omic integration method that models the cascading effects of genetic variants from epigenome to transcriptome and eventually to the phenome in identifying target genes influenced by risk alleles. This cascading epigenomic analysis for GWAS, which we refer to as CEWAS, comprises two types of models: one for linking cis genetic effects to epigenomic variation and another for linking cis epigenomic variation to gene expression. Applying these models in cascade to GWAS summary statistics generates gene level statistics that reflect genetically-driven epigenomic effects. We show on sixteen brain-related GWAS that CEWAS provides higher gene detection rate than related methods, and finds disease relevant genes and gene sets that point toward less explored biological processes. CEWAS thus presents a novel means for exploring the regulatory landscape of GWAS variants in uncovering disease mechanisms. SummaryThe majority of genetic variants detected in genome wide association studies (GWAS) exert their effects on phenotypes through gene regulation. Motivated by this observation, we propose a multi-omic integration method that models the cascading effects of genetic variants from epigenome to transcriptome and eventually to the phenome in identifying target genes influenced by risk alleles. This cascading epigenomic analysis for GWAS, which we refer to as CEWAS, combines the effect of genetic variants on DNA methylation as well as gene expression. We show on sixteen brain-related GWAS that CEWAS provides higher gene detection rate than related methods, and finds disease relevant genes and gene sets that point toward less explored biological processes.

bioinformatics↗

bioRxiv2019

Gene expression profiles complement the analysis of genomic modifiers of the clinical onset of Huntington disease

Huntington disease (HD) is a neurodegenerative disorder that is caused by a CAG repeat expansion in the HTT gene. In an attempt to identify genomic modifiers that contribute towards the age of onset of HD, we performed a transcriptome wide association study assessing heritable differences in genetically determined expression in diverse tissues, employing genome wide data from over 4,000 patients. This identified genes that showed evidence for colocalization and replication, with downstream functional validation being performed in isogenic HD stem cells and patient brains. Enrichment analyses detected associations with various biologically-relevant gene sets and striatal coexpression modules that are mediated by CAG length. Further, cortical coexpression modules that are relevant for HD onset were also associated with cognitive decline and HD-related traits in a longitudinal cohort. In summary, the combination of population-scale gene expression information with HD patient genomic data identified novel modifier genes for the disorder.

genomics↗

bioRxiv2019

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 Alzheimers 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), and provide the results as a community resource. We find substantially improved power in the meta-analysis over individual cohort analyses, particularly in comparison to the Genotype-Tissue Expression (GTEx) Project eQTL. In addition, we observed differences in eQTL patterns between cerebral and cerebellar brain regions. We provide these brain eQTL as a common resource for use across the community in research programs. 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).

genetics↗

bioRxiv2019

Accurate ethnicity prediction from placental DNA methylation data

BackgroundThe influence of genetics on variation in DNA methylation (DNAme) is well documented. Yet confounding from population stratification is often unaccounted for in DNAme association studies. Existing approaches to address confounding by population stratification using DNAme data may not generalize to populations or tissues outside those in which they were developed. To aid future placental DNAme studies in assessing population stratification, we developed an ethnicity classifier, PlaNET (Placental DNAme Elastic Net Ethnicity Tool), using five cohorts with Infinium Human Methylation 450k BeadChip array (HM450k) data from placental samples that is also compatible with the newer EPIC platform.\n\nResultsData from 509 placental samples was used to develop PlaNET and show that it accurately predicts (accuracy = 0.938, kappa = 0.823) major classes of self-reported ethnicity/race (African: n = 58, Asian: n = 53, Caucasian: n = 389), and produces ethnicity probabilities that are highly correlated with genetic ancestry inferred from genome-wide SNP arrays (>2.5 million SNP) and ancestry informative markers (n = 50 SNPs). PlaNETs ethnicity classification relies on 1860 HM450K microarray sites, and over half of these were linked to nearby genetic polymorphisms (n = 955). Our placental-optimized method outperforms existing approaches in assessing population stratification in placental samples from individuals of Asian, African, and Caucasian ethnicities.\n\nConclusionPlaNET provides an improved approach to address population stratification in placental DNAme association studies. The method can be applied to predict ethnicity as a discrete or continuous variable and will be especially useful when self-reported ethnicity information is missing and genotyping markers are unavailable. PlaNET is available as an R package at (https://github.com/wvictor14/planet).

genomics↗

bioRxiv2019

Deconvolving the contributions of cell-type heterogeneity on cortical gene expression

Complexity of cell-type composition has created much skepticism surrounding the interpretation of brain bulk-tissue transcriptomic studies. We generated paired tissue genome-wide gene expression data and immunohistochemistry data, enabling us to assess statistical methods for modeling and estimating cellular heterogeneity in the brain. We demonstrate that several algorithms that rely on single-cell and cell-sorted data to define cell marker gene sets yield accurate relative and absolute estimates of constituent cell-type proportions.

neuroscience↗

bioRxiv2019

Meta-analysis of the human brain transcriptome identifies heterogeneity across human AD coexpression modules robust to sample collection and methodological approach

Alzheimers disease (AD) is a complex and heterogenous brain disease that affects multiple inter-related biological processes. This complexity contributes, in part, to existing difficulties in the identification of successful disease-modifying therapeutic strategies. To address this, systems approaches are being used to characterize AD-related disruption in molecular state. To evaluate the consistency across these molecular models, a consensus atlas of the human brain transcriptome was developed through coexpression meta-analysis across the AMP-AD consortium. Consensus analysis was performed across five coexpression methods used to analyze RNA-seq data collected from 2114 samples across 7 brain regions and 3 research studies. From this analysis, five consensus clusters were identified that described the major sources of AD-related alterations in transcriptional state that were consistent across studies, methods, and samples. AD genetic associations, previously studied AD-related biological processes, and AD targets under active investigation were enriched in only three of these five clusters. The remaining two clusters demonstrated strong heterogeneity between males and females in AD-related expression that was consistently observed across studies. AD transcriptional modules identified by systems analysis of individual AMP-AD teams were all represented in one of these five consensus clusters except ROS/MAP-identified Module 109, which was specific for genes that showed the strongest association with changes in AD-related gene expression across consensus clusters. The other two AMP-AD transcriptional analyses reported modules that were enriched in one of the two sex-specific Consensus Clusters. The fifth cluster has not been previously identified and was enriched for genes related to proteostasis. This study provides an atlas to map across biological inquiries of AD with the goal of supporting an expansion in AD target discovery efforts.

systems biology↗

bioRxiv2018

Genetic variation in the Major Histocompatibility Complex and association with depression

BackgroundThe prevalence of depression is higher in individuals suffering from autoimmune diseases, but the mechanisms underlying the observed comorbidities are unknown. Epidemiological findings point to a bi-directional relationship - that depression increases the risk of developing an autoimmune disease, and vice-versa. Shared genetic etiology is a plausible explanation for the overlap between depression and autoimmune diseases. In this study we tested whether genetic variation in the Major Histocompatibility Complex (MHC), which is associated with risk for autoimmune diseases, is also associated with risk for depression.\n\nMethodWe fine-mapped the classical MHC (chr6: 29.6-33.1 Mb), imputing 216 Human Leukocyte Antigen (HLA) alleles and four Complement Component 4 (C4) haplotypes in studies from the Psychiatric Genomics Consortium (PGC) Major Depressive Disorder (MDD) working group and the UK Biobank (UKB). In the 26 PGC-MDD studies, cases met a lifetime diagnosis of MDD, determined by a structured diagnostic interview. In the UKB, cases and controls were identified from an online mental health questionnaire. The total sample size was 45,149 depression cases and 86,698 controls. We tested for association between depression status and imputed MHC variants in each study and performed an inverse-variance weighted meta-analysis across the PGC-MDD and UKB samples, applying both a conservative region-wide significance threshold (3.9-e6) and a candidate threshold (1.6e-4).\n\nResultsNo HLA alleles or C4 haplotypes were associated with depression at the conservative threshold in the PGC, UKB or meta-analysis. HLA-B*08:01 was associated with modest protection for depression at the candidate threshold in the meta-analysis. Under the conservative threshold, 70 SNPs were detected in the UKB and 143 SNPs were detected in the meta-analysis, mirroring previous findings from highly powered GWAS of depression.\n\nDiscussionWe found no evidence that HLA alleles, which play a major role in the genetic susceptibility to autoimmune diseases, or C4 haplotypes, which are strongly associated with schizophrenia, confer risk for depression. These results indicate that autoimmune diseases and depression do not share common risk loci of moderate or large effect in the MHC.

genetics↗

bioRxiv2018

Differences in DNA methylation of white blood cell types at birth and in adulthood reflect postnatal immune maturation and influence accuracy of cell type prediction

BackgroundDNA methylation profiling of peripheral blood leukocytes has many research applications, and characterizing the changes in DNA methylation of specific white blood cell types between newborn and adult could add insight into the maturation of the immune system. As a consequence of developmental changes, DNA methylation profiles derived from adult white blood cells are poor references for prediction of cord blood cell types from DNA methylation data. We thus examined cell-type specific differences in DNA methylation in leukocyte subsets between cord and adult blood, and assessed the impact of these differences on prediction of cell types in cord blood.\n\nResultsThough all cell types showed differences between cord and adult blood, some specific patterns stood out that reflected how the immune system changes after birth. In cord blood, lymphoid cells showed less variability than in adult, potentially demonstrating their naive status. In fact, cord CD4 and CD8 T cells were so similar that genetic effects on DNA methylation were greater than cell type effects in our analysis, and CD8 T cell frequencies remained difficult to predict, even after optimizing the library used for cord blood composition estimation. Myeloid cells showed fewer changes between cord and adult and also less variability, with monocytes showing the fewest sites of DNA methylation change between cord and adult. Finally, including nucleated red blood cells in the reference library was necessary for accurate cell type predictions in cord blood.\n\nConclusionChanges in DNA methylation with age were highly cell type specific, and those differences paralleled what is known about the maturation of the postnatal immune system.

genetics↗

bioRxiv2018

DNA Methylation Network Estimation with Sparse Latent Gaussian Graphical Model

Inferring molecular interaction networks from genomics data is important for advancing our understanding of biological processes. Whereas considerable research effort has been placed on inferring such networks from gene expression data, network estimation from DNA methylation data has received very little attention due to the substantially higher dimensionality and complications with result interpretation for non-genic regions. To combat these challenges, we propose here an approach based on sparse latent Gaussian graphical model (SLGGM). The core idea is to perform network estimation on q latent variables as opposed to d CpG sites, with q<<d. To impose a correspondence between the latent variables and genes, we use the distance between CpG sites and transcription starting sites of the genes to generate a prior on the CpG sites latent class membership. We evaluate this approach on synthetic data, and show on real data that the gene network estimated from DNA methylation data significantly explains gene expression patterns in unseen datasets.

genomics↗

bioRxiv2018

Epigenome-wide study uncovers tau pathology-driven changes of chromatin organization in the aging human brain

Accumulation of tau and amyloid-{beta} are two pathologic hallmarks of Alzheimers disease (AD). Here, we conducted an epigenome-wide association study using the H3K9 acetylation (H3K9Ac) mark in 669 aged human prefrontal cortices: in contrast to amyloid-{beta}, tau protein burden had a broad effect on the epigenome, affecting 5,590 out of 26,384 H3K9Ac domains. Tau-related alterations aggregated in large genomic segments reflecting spatial chromatin organization, and the magnitude of these effects correlated with the segments nuclear lamina association. We confirmed the functional relevance of these chromatin changes by demonstrating (1) consistent transcriptional changes in three independent datasets and (2) similar findings in two AD mouse models. Finally, we found that tau overexpression in iPSC-derived neurons disrupted chromatin organization and that these effects could be blocked by a small molecule predicted to reverse the tau effect. Thus, we report large-scale tau-driven chromatin rearrangements in the aging human brain that may be reversible with HSP90 inhibitors.

neuroscience↗

bioRxiv2018

A multi-omic atlas of the human frontal cortex for aging and Alzheimer's disease research

We initiated the systematic profiling of the dorsolateral prefrontal cortex obtained from a subset of autopsied individuals enrolled in the Religious Orders Study (ROS) or the Rush Memory and Aging Project (MAP), which are jointly designed and belong to a very few prospective studies of aging and dementia with detailed, longitudinal cognitive phenotyping during life and a quantitative, structured neuropathologic examination after death for >3,322 subjects. Here, we outline the first generation of data including genome-wide genotypes (n=2,090), whole genome sequencing (n=1,179), DNA methylation (n=740), chromatin immunoprecipitation with sequencing using an anti-Histone 3 Lysine 9 acetylation (H3K9Ac) antibody (n=712), RNA sequencing (n=638), and miRNA profile (n=702). Generation of other omic data including ATACseq, proteomic and metabolomics profiles is ongoing. Thanks to its prospective design and recruitment of older, non-demented individuals, these data can be repurposed to investigate a large number of syndromic and quantitative neuroscience phenotypes. The many subjects that are cognitively non-impaired at death also offer insights into the biology of the human brain in older non-impaired individuals.

cell biology↗