Biology subjects

Eskin, E.

Publications and source records attributed to Eskin, E..

At least 37 records · Page 2Linked to original sources

bioRxiv2020

A comprehensive benchmarking of WGS-based structural variant callers

Advances in whole genome sequencing promise to enable the accurate and comprehensive structural variant (SV) discovery. Dissecting SVs from whole genome sequencing (WGS) data presents a substantial number of challenges and a plethora of SV-detection methods have been developed. Currently, there is a paucity of evidence which investigators can use to select appropriate SV-detection tools. In this paper, we evaluated the performance of SV-detection tools using a comprehensive PCR-confirmed gold standard set of SVs. In contrast to the previous benchmarking studies, our gold standard dataset included a complete set of SVs allowing us to report both precision and sensitivity rates of SV-detection methods. Our study investigates the ability of the methods to detect deletions, thus providing an optimistic estimate of SV detection performance, as the SV-detection methods that fail to detect deletions are likely to miss more complex SVs. We found that SV-detection tools varied widely in their performance, with several methods providing a good balance between sensitivity and precision. Additionally, we have determined the SV callers best suited for low and ultra-low pass sequencing data.

bioinformatics↗

bioRxiv2020

Annotating Gene Ontology terms for protein sequences with the Transformer model

Predicting functions for novel amino acid sequences is a long-standing research problem. The Uniprot database which contains protein sequences annotated with Gene Ontology (GO) terms, is one commonly used training dataset for this problem. Predicting protein functions can then be viewed as a multi-label classification problem where the input is an amino acid sequence and the output is a set of GO terms. Recently, deep convolutional neural network (CNN) models have been introduced to annotate GO terms for protein sequences. However, the CNN architecture can only model close-range interactions between amino acids in a sequence. In this paper, first, we build a novel GO annotation model based on the Transformer neural network. Unlike the CNN architecture, the Transformer models all pairwise interactions for the amino acids within a sequence, and so can capture more relevant information from the sequences. Indeed, we show that our adaptation of Transformer yields higher classification accuracy when compared to the recent CNN-based method DeepGO. Second, we modify our model to take motifs in the protein sequences found by BLAST as additional input features. Our strategy is different from other ensemble approaches that average the outcomes of BLAST-based and machine learning predictors. Third, we integrate into our Transformer the metadata about the protein sequences such as 3D structure and protein-protein interaction (PPI) data. We show that such information can greatly improve the prediction accuracy, especially for rare GO labels.

bioinformatics↗

bioRxiv2020

Metalign: Efficient alignment-based metagenomic profiling via containment min hash

Whole-genome shotgun sequencing enables the analysis of microbial communities in unprecedented detail, with major implications in medicine and ecology. Predicting the presence and relative abundances of microbes in a sample, known as "metagenomic profiling", is a critical first step in microbiome analysis. Existing profiling methods have been shown to suffer from poor false positive or false negative rates, while alignment-based approaches are often considered accurate but computationally infeasible. Here we present a novel method, Metalign, that addresses these concerns by performing efficient alignment-based metagenomic profiling. We use a containment min hash approach to reduce the reference database size dramatically before alignment and a method to estimate organism relative abundances in the sample by resolving reads aligned to multiple genomes. We show that Metalign achieves significantly improved results over existing methods on simulated datasets from a large benchmarking study, CAMI, and performs well on in vitro mock community data and environmental data from the Tara Oceans project. Metalign is freely available at https://github.com/nlapier2/Metalign, along with the results and plots used in this paper, and a docker image is also available at https://hub.docker.com/repository/docker/nlapier2/metalign.

bioinformatics↗

bioRxiv2020

Identifying Causal Variants by Fine Mapping Across Multiple Studies

Increasingly large Genome-Wide Association Studies (GWAS) have yielded numerous variants associated with many complex traits, motivating the development of "fine mapping" methods to identify which of the associated variants are causal. Additionally, GWAS of the same trait for different populations are increasingly available, raising the possibility of refining fine mapping results further by leveraging different linkage disequilibrium (LD) structures across studies. Here, we introduce multiple study causal variants identification in associated regions (MsCAVIAR), a method that extends the popular CAVIAR fine mapping framework to a multiple study setting using a random effects model. MsCAVIAR only requires summary statistics and LD as input, accounts for uncertainty in association statistics using a multivariate normal model, allows for multiple causal variants at a locus, and explicitly models the possibility of different SNP effect sizes in different populations. In a trans-ethnic, trans-biobank Type 2 Diabetes analysis, we show that MsCAVIAR returns causal set sizes that are over 20% smaller than those given by current state of the art methods for trans-ethnic fine-mapping.

bioinformatics↗

bioRxiv2019

Accurate modeling of replication rates in genome-wide association studies by accounting for winner's curse and study-specific heterogeneity

Genome-wide association studies (GWAS) have identified thousands of genetic variants associated with complex human traits, but only a fraction of variants identified in discovery studies achieve significance in replication studies. Replication in GWAS has been well-studied in the context of Winners Curse, which is the inflation of effect size estimates for significant variants due to statistical chance. However, Winners Curse is often not sufficient to explain lack of replication. Another reason why studies fail to replicate is that there are fundamental differences between the discovery and replication studies. A confounding factor can create the appearance of a significant finding while actually being an artifact that will not replicate in future studies. We propose a statistical framework that utilizes GWAS and replication studies to jointly model Winners Curse and study-specific heterogeneity due to confounding factors. We apply this framework to 100 GWAS from the Human GWAS Catalog and observe that there is a large range in the level of estimated confounding. We demonstrate how this framework can be used to distinguish when studies fail to replicate due to statistical noise and when they fail due to confounding.

bioinformatics↗

bioRxiv2019

Learning Representations for Gene Ontology Terms by Contextualized Text Encoders

Protein functions can be described by the Gene Ontology (GO) terms, allowing us to compare the functions of two proteins by measuring the similarity of the terms assigned to them. Recent works have applied neural network models to derive the vector representations for GO terms and compute similarity scores for these terms by comparing their vector embeddings. There are two typical ways to embed GO terms into vectors; a model can either embed the definitions of the terms or the topology of the terms in the ontology. In this paper, we design three tasks to critically evaluate the GO embeddings of two recent neural network models, and further introduce additional models for embedding GO terms, adapted from three popular neural network frameworks: Graph Convolution Network (GCN), Embeddings from Language Models (ELMo), and Bidirectional Encoder Representations from Transformers (BERT), which have not yet been explored in previous works. Task 1 studies edge cases where the GO embeddings may not provide meaningful similarity scores for GO terms. We find that all neural network based methods fail to produce high similarity scores for related terms when these terms have low Information Content values. Task 2 is a canonical task which estimates how well GO embeddings can compare functions of two orthologous genes or two interacting proteins. The best neural network methods for this task are those that embed GO terms using their definitions, and the differences among such methods are small. Task 3 evaluates how GO embeddings affect the performance of GO annotation methods, which predict whether a protein should be labeled by certain GO terms. When the annotation datasets contain many samples for each GO label, GO embeddings do not improve the classification accuracy. Machine learning GO annotation methods often remove rare GO labels from the training datasets so that the model parameters can be efficiently trained. We evaluate whether GO embeddings can improve prediction of rare labels unseen in the training datasets, and find that GO embeddings based on the BERT framework achieve the best results in this setting. We present our embedding methods and three evaluation tasks as the basis for future research on this topic.

bioinformatics↗

bioRxiv2019

Large-scale Genetic Analysis Identifies 66 Novel Loci for Asthma

We carried out a genome-wide association study (GWAS) for asthma in UK Biobank, followed by a meta-analysis with results from the Trans-National Asthma Genetic Consortium (TAGC). 66 novel genomic regions were identified, bringing the number of known asthma susceptibility loci to 211. Significant gene-sex interactions were also observed where susceptibility alleles, either individually or as a function of polygenic risk scores, increased asthma risk to a greater extent in men than women. Bioinformatics analyses demonstrated that asthma-associated variants were enriched for colocalizing to regions of open chromatic in immune cells and identified candidate causal genes at 52 of the novel loci, including CD52. An anti-CD52 (-CD52) antibody mimicked the immune cell-depleting effects of an FDA-approved human -CD52 antibody and reduced allergen-induced airway hyperreactivity in mice. These results further elucidate the genetic architecture of asthma, provide evidence that the immune system plays a prominent role in its pathogenesis, and suggest that CD52 represents a potentially novel therapeutic target for treating asthma.

genetics↗

bioRxiv2018

Analysis of multiple fungal sequence repositories highlights shortcomings in microbial databases

Metagenomics studies leverage genomic reference databases to generate discoveries in basic science and translational research. However, current microbial studies use disparate reference databases that lack consistent standards of specimen inclusion, data preparation, taxon labelling and accessibility, hindering their quality and comprehensiveness, and calling for the establishment of recommendations for reference genome database assembly. Here, we analyze existing fungal and bacterial databases and discuss guidelines for the development of a master reference database that promises to improve the quality and quantity of omics research.

bioinformatics↗

bioRxiv2018

Leveraging allelic heterogeneity to increase power of association testing

The standard genome-wide association studies (GWAS) detects an association between a single variant and a phenotype of interest. Recently, several studies reported that at many risk loci, there may exist multiple causal variants. For a locus with multiple causal variants with small effect sizes, the standard association test is underpowered to detect the associations. Alternatively, an approach considering effects of multiple variants simultaneously may increase statistical power by leveraging effects of multiple causal variants. In this paper, we propose a new statistical method, Model-based Association test Reflecting causal Status (MARS), that tries to find an association between variants in risk loci and a phenotype, considering the causal status of the variants. One of the main advantages of MARS is that it only requires the existing summary statistics to detect associated risk loci. Thus, MARS is applicable to any association study with summary statistics, even though individual level data is not available for the study. Utilizing extensive simulated data sets, we show that MARS increases the power of detecting true associated risk loci compared to previous approaches that consider multiple variants, while robustly controls the type I error. Applied to data of 44 tissues provided by the Genotype-Tissue Expression (GTEx) consortium, we show that MARS identifies more eGenes compared to previous approaches in most of the tissues; e.g. MARS identified 16% more eGenes than the ones reported by the GTEx consortium. Moreover, applied to Northern Finland Birth Cohort (NFBC) data, we demonstrate that MARS effectively identifies association loci with improved power (56% of more loci found by MARS) inGWAS studies compared to the standard association test.

genetics↗

bioRxiv2018

A comprehensive analysis of the usability and archival stability of omics computational tools and resources

Developing new software tools for analysis of large-scale biological data is a key component of advancing modern biomedical research. Scientific reproduction of published findings requires running computational tools on data generated by such studies, yet little attention is presently allocated to the installability and archival stability of computational software tools. Scientific journals require data and code sharing, but none currently require authors to guarantee the continuing functionality of newly published tools. We have estimated the archival stability of computational biology software tools by performing an empirical analysis of the internet presence for 36,702 omics software resources published from 2005 to 2017. We found that almost 28% of all resources are currently not accessible through URLs published in the paper they first appeared in. Among the 98 software tools selected for our installability test, 51% were deemed \"easy to install,\" and 28% of the tools failed to be installed at all due to problems in the implementation. Moreover, for papers introducing new software, we found that the number of citations significantly increased when authors provided an easy installation process. We propose for incorporation into journal policy several practical solutions for increasing the widespread installability and archival stability of published bioinformatics software.

bioinformatics↗

bioRxiv2018

Cell-type-specific resolution epigenetics without the need for cell sorting or single-cell biology

High costs and technical limitations of cell sorting and single-cell techniques currently restrict the collection of large-scale, cell-type-specific DNA methylation data. This, in turn, impedes our ability to tackle key biological questions that pertain to variation within a population, such as identification of disease-associated genes at a cell-type-specific resolution. Here, we show mathematically and empirically that cell-type-specific methylation levels of an individual can be learned from its tissue-level bulk data, conceptually emulating the case where the individual has been profiled with a single-cell resolution and then signals were aggregated in each cell population separately. Provided with this unprecedented way to perform powerful large-scale epigenetic studies with cell-type-specific resolution, we revisit previous studies with tissue-level bulk methylation and reveal novel associations with leukocyte composition in blood and with rheumatoid arthritis. For the latter, we further show consistency with validation data collected from sorted leukocyte sub-types. Corresponding software is available from: https://github.com/cozygene/TCA.

bioinformatics↗

bioRxiv2018

A unifying framework for summary statistic imputation

Imputation has been widely utilized to aid and interpret the results of Genome-Wide Association Studies(GWAS). Imputation can increase the power to identify associations when the causal variant was not directly observed or typed in the GWAS. There are two broad classes of methods for imputation. The first class imputes the genotypes at the untyped variants given the genotypes at the typed variants and then performs a statistical test of association at the imputed variants. The second class of methods, summary statistic imputation, directly imputes the association statics at the untyped variants given the association statistics observed at the typed variants. This second class of methods is appealing as it tends to be computationally efficient while only requiring the summary statistics from a study while the former class requires access to individual-level data that can be difficult to obtain. The statistical properties of these two classes of imputation methods have not been fully understood. In this paper, we show that the two classes of imputation methods are equivalent, i.e., have identical asymptotic multivariate normal distributions with zero mean and minor variations in the covariance matrix, under some reasonable assumptions. Using this equivalence, we can understand the effect of imputation methods on power. We show that a commonly employed modification of summary statistic imputation that we term summary statistic imputation with variance re-weighting generally leads to a loss in power. On the other hand, our proposed method, summary statistic imputation without performing variance re-weighting, fully accounts for imputation uncertainty while achieving better power.

bioinformatics↗

bioRxiv2018

Leveraging allele-specific expression to refine fine-mapping for eQTL studies

Many disease risk loci identified in genome-wide association studies are present in non-coding regions of the genome. It is hypothesized that these variants affect complex traits by acting as expression quantitative trait loci (eQTLs) that influence expression of nearby genes. This indicates that many causal variants for complex traits are likely to be causal variants for gene expression. Hence, identifying causal variants for gene expression is important for elucidating the genetic basis of not only gene expression but also complex traits. However, detecting causal variants is challenging due to complex genetic correlation among variants known as linkage disequilibrium (LD) and the presence of multiple causal variants within a locus. Although several fine-mapping approaches have been developed to overcome these challenges, they may produce large sets of putative causal variants when true causal variants are in high LD with many non-causal variants. In eQTL studies, there is an additional source of information that can be used to improve fine-mapping called allele-specific expression (ASE) that measures imbalance in gene expression due to different alleles. In this work, we develop a novel statistical method that leverages both ASE and eQTL information to detect causal variants that regulate gene expression. We illustrate through simulations and application to the Genotype-Tissue Expression (GTEx) dataset that our method identifies the true causal variants with higher specificity than an approach that uses only eQTL information. In the GTEx dataset, our method achieves the median reduction rate of 11% in the number of putative causal variants.\n\nContactJaeHoonSul@mednet.ucla.edu, eeskin@cs.ucla.edu

genetics↗

bioRxiv2018

MiCoP: Microbial Community Profiling method for detecting viral and fungal organisms in metagenomic samples

BackgroundHigh throughput sequencing has spurred the development of metagenomics, which involves the direct analysis of microbial communities in various environments such as soil, ocean water, and the human body. Many existing methods based on marker genes or k-mers have limited sensitivity or are too computationally demanding for many users. Additionally, most work in metagenomics has focused on bacteria and archaea, neglecting to study other key microbes such as viruses and eukaryotes.\n\nResultsHere we present a method, MiCoP (Microbiome Community Profiling), that uses fast-mapping of reads to build a comprehensive reference database of full genomes from viruses and eukaryotes to achieve maximum read usage and enable the analysis of the virome and eukaryome in each sample. We demonstrate that mapping of metagenomic reads is feasible for the smaller viral and eukaryotic reference databases. We show that our method is accurate on simulated and mock community data and identifies many more viral and fungal species than previously-reported results on real data from the Human Microbiome Project.\n\nConclusionsMiCoP is a mapping-based method that proves more effective than existing methods at abundance profiling of viruses and eukaryotes in metagenomic samples. MiCoP can be used to detect the full diversity of these communities. The code, data, and documentation is publicly available on GitHub at: https://github.com/smangul1/MiCoP

bioinformatics↗

bioRxiv2017

The Multivariate Normal Distribution Framework for Analyzing Association Studies

Genome-wide association studies (GWAS) have discovered thousands of variants involved in common human diseases. In these studies, frequencies of genetic variants are compared between a cohort of individuals with a disease (cases) and a cohort of healthy individuals (controls). Any variant that has a significantly different frequency between the two cohorts is considered an associated variant. A challenge in the analysis of GWAS studies is the fact that human population history causes nearby genetic variants in the genome to be correlated with each other. In this review, we demonstrate how to utilize the multivariate normal (MVN) distribution to explicitly take into account the correlation between genetic variants in a comprehensive framework for analysis of GWAS. We show how the MVN framework can be applied to perform association testing, correct for multiple hypothesis testing, estimate statistical power, and perform fine mapping and imputation.

genetics↗

bioRxiv2017

Leveraging molecular QTL to understand the genetic architecture of diseases and complex traits

There is increasing evidence that many GWAS risk loci are molecular QTL for gene ex-pression (eQTL), histone modification (hQTL), splicing (sQTL), and/or DNA methylation (meQTL). Here, we introduce a new set of functional annotations based on causal posterior prob-abilities (CPP) of fine-mapped molecular cis-QTL, using data from the GTEx and BLUEPRINT consortia. We show that these annotations are very strongly enriched for disease heritability across 41 independent diseases and complex traits (average N = 320K): 5.84x for GTEx eQTL, and 5.44x for eQTL, 4.27-4.28x for hQTL (H3K27ac and H3K4me1), 3.61x for sQTL and 2.81x for meQTL in BLUEPRINT (all P [≤] 1.39e-10), far higher than enrichments obtained using stan-dard functional annotations that include all significant molecular cis-QTL (1.17-1.80x). eQTL annotations that were obtained by meta-analyzing all 44 GTEx tissues generally performed best, but tissue-specific blood eQTL annotations produced stronger enrichments for autoimmune dis-eases and blood cell traits and tissue-specific brain eQTL annotations produced stronger enrich-ments for brain-related diseases and traits, despite high cis-genetic correlations of eQTL effect sizes across tissues. Notably, eQTL annotations restricted to loss-of-function intolerant genes from ExAC were even more strongly enriched for disease heritability (17.09x; vs. 5.84x for all genes; P = 4.90e-17 for difference). All molecular QTL except sQTL remained significantly enriched for disease heritability in a joint analysis conditioned on each other and on a broad set of functional annotations from previous studies, implying that each of these annotations is uniquely informative for disease and complex trait architectures.

genetics↗

bioRxiv2017

Addressing the digital divide in contemporary biology: Lessons from teaching UNIX

Researchers in the biomedical sciences increasingly rely on applications that lack a graphical interface and require inputting code that, such as UNIX. Scientists who are not trained in computer science face an enormous challenge in analyzing the high-throughput data their research groups generate. We present a training model for use of command-line tools when the learner has little to no prior knowledge of UNIX.

scientific communication and education↗

bioRxiv2017

A Bayesian Framework for Estimating Cell Type Composition from DNA Methylation Without the Need for Methylation Reference

We introduce a Bayesian semi-supervised method for estimating cell counts from DNA methylation by leveraging an easily obtainable prior knowledge on the cell type composition distribution of the studied tissue. We show mathematically and empirically that alternative methods which attempt to infer explicit cell counts without methylation reference can only capture linear combinations of cell counts rather than provide one component per cell type. Our approach, which allows the construction of a set of components such that each component corresponds to a single cell type, therefore provides a new opportunity to investigate cell compositions in genomic studies of tissues for which it was not possible before.

bioinformatics↗