NPSV-deep: a deep learning method for genotyping structural variants in short read genome sequencing data.

MOTIVATION: Structural variants (SVs) play a causal role in numerous diseases but can be difficult to detect and accurately genotype (determine zygosity) with short-read genome sequencing data (SRS). Improving SV genotyping accuracy in SRS data, particularly for the many SVs first detected with long-read sequencing, will improve our understanding of genetic variation. RESULTS: NPSV-deep is a deep learning-based approach for genotyping previously reported insertion and deletion SVs that recasts this task as an image similarity problem. NPSV-deep predicts the SV genotype based on the similarity between pileup images generated from the actual SRS data and matching SRS simulations. We show that NPSV-deep consistently matches or improves upon the state-of-the-art for SV genotyping accuracy across different SV call sets, samples and variant types, including a 25% reduction in genotyping errors for the Genome-in-a-Bottle (GIAB) high-confidence SVs. NPSV-deep is not limited to the SVs as described; it improves deletion genotyping concordance a further 1.5 percentage points for GIAB SVs (92%) by automatically correcting imprecise/incorrectly described SVs. AVAILABILITY AND IMPLEMENTATION: Python/C++ source code and pre-trained models freely available at https://github.com/mlinderm/npsv2.

NPSV: A simulation-driven approach to genotyping structural variants in whole-genome sequencing data.

BACKGROUND: Structural variants (SVs) play a causal role in numerous diseases but are difficult to detect and accurately genotype (determine zygosity) in whole-genome next-generation sequencing data. SV genotypers that assume that the aligned sequencing data uniformly reflect the underlying SV or use existing SV call sets as training data can only partially account for variant and sample-specific biases. RESULTS: We introduce NPSV, a machine learning-based approach for genotyping previously discovered SVs that uses next-generation sequencing simulation to model the combined effects of the genomic region, sequencer, and alignment pipeline on the observed SV evidence. We evaluate NPSV alongside existing SV genotypers on multiple benchmark call sets. We show that NPSV consistently achieves or exceeds state-of-the-art genotyping accuracy across SV call sets, samples, and variant types. NPSV can specifically identify putative de novo SVs in a trio context and is robust to offset SV breakpoints. CONCLUSIONS: Growing SV databases and the increasing availability of SV calls from long-read sequencing make stand-alone genotyping of previously identified SVs an increasingly important component of genome analyses. By treating potential biases as a "black box" that can be simulated, NPSV provides a framework for accurately genotyping a broad range of SVs in both targeted and genome-scale applications.

Development and Validation of a Comprehensive Genomics Knowledge Scale.

BACKGROUND: Genomic testing is increasingly employed in clinical, research, educational, and commercial contexts. Genomic literacy is a prerequisite for the effective application of genomic testing, creating a corresponding need for validated tools to assess genomics knowledge. We sought to develop a reliable measure of genomics knowledge that incorporates modern genomic technologies and is informative for individuals with diverse backgrounds, including those with clinical/life sciences training. METHODS: We developed the GKnowM Genomics Knowledge Scale to assess the knowledge needed to make an informed decision for genomic testing, appropriately apply genomic technologies and participate in civic decision-making. We administered the 30-item draft measure to a calibration cohort (n = 1,234) and subsequent participants to create a combined validation cohort (n = 2,405). We performed a multistage psychometric calibration and validation using classical test theory and item response theory (IRT) and conducted a post-hoc simulation study to evaluate the suitability of a computerized adaptive testing (CAT) implementation. RESULTS: Based on exploratory factor analysis, we removed 4 of the 30 draft items. The resulting 26-item GKnowM measure has a single dominant factor. The scale internal consistency is α = 0.85, and the IRT 3-PL model demonstrated good overall and item fit. Validity is demonstrated with significant correlation (r = 0.61) with an existing genomics knowledge measure and significantly higher scores for individuals with adequate health literacy and healthcare providers (HCPs), including HCPs who work with genomic testing. The item bank is well suited to CAT, achieving high accuracy (r = 0.97 with the full measure) while administering a mean of 13.5 items. CONCLUSION: GKnowM is an updated, broadly relevant, rigorously validated 26-item measure for assessing genomics knowledge that we anticipate will be useful for assessing population genomic literacy and evaluating the effectiveness of genomics educational interventions.

MySeq: privacy-protecting browser-based personal Genome analysis for genomics education and exploration.

BACKGROUND: The complexity of genome informatics is a recurring challenge for genome exploration and analysis by students and other non-experts. This complexity creates a barrier to wider implementation of experiential genomics education, even in settings with substantial computational resources and expertise. Reducing the need for specialized software tools will increase access to hands-on genomics pedagogy. RESULTS: MySeq is a React.js single-page web application for privacy-protecting interactive personal genome analysis. All analyses are performed entirely in the user's web browser eliminating the need to install and use specialized software tools or to upload sensitive data to an external web service. MySeq leverages Tabix-indexing to efficiently query whole genome-scale variant call format (VCF) files stored locally or available remotely via HTTP(s) without loading the entire file. MySeq currently implements variant querying and annotation, physical trait prediction, pharmacogenomic, polygenic disease risk and ancestry analyses to provide representative pedagogical examples; and can be readily extended with new analysis or visualization components. CONCLUSIONS: MySeq supports multiple pedagogical approaches including independent exploration and interactive online tutorials. MySeq has been successfully employed in an undergraduate human genome analysis course where it reduced the barriers-to-entry for hands-on human genome analysis.

DECA: scalable XHMM exome copy-number variant calling with ADAM and Apache Spark.

BACKGROUND: XHMM is a widely used tool for copy-number variant (CNV) discovery from whole exome sequencing data but can require hours to days to run for large cohorts. A more scalable implementation would reduce the need for specialized computational resources and enable increased exploration of the configuration parameter space to obtain the best possible results. RESULTS: DECA is a horizontally scalable implementation of the XHMM algorithm using the ADAM framework and Apache Spark that incorporates novel algorithmic optimizations to eliminate unneeded computation. DECA parallelizes XHMM on both multi-core shared memory computers and large shared-nothing Spark clusters. We performed CNV discovery from the read-depth matrix in 2535 exomes in 9.3 min on a 16-core workstation (35.3× speedup vs. XHMM), 12.7 min using 10 executor cores on a Spark cluster (18.8× speedup vs. XHMM), and 9.8 min using 32 executor cores on Amazon AWS' Elastic MapReduce. We performed CNV discovery from the original BAM files in 292 min using 640 executor cores on a Spark cluster. CONCLUSIONS: We describe DECA's performance, our algorithmic and implementation enhancements to XHMM to obtain that performance, and our lessons learned porting a complex genome analysis application to ADAM and Spark. ADAM and Apache Spark are a performant and productive platform for implementing large-scale genome analyses, but efficiently utilizing large clusters can require algorithmic optimizations and careful attention to Spark's configuration parameters.

Predispositional genome sequencing in healthy adults: design, participant characteristics, and early outcomes of the PeopleSeq Consortium.

BACKGROUND: Increasing numbers of healthy individuals are undergoing predispositional personal genome sequencing. Here we describe the design and early outcomes of the PeopleSeq Consortium, a multi-cohort collaboration of predispositional genome sequencing projects, which is examining the medical, behavioral, and economic outcomes of returning genomic sequencing information to healthy individuals. METHODS: Apparently healthy adults who participated in four of the sequencing projects in the Consortium were included. Web-based surveys were administered before and after genomic results disclosure, or in some cases only after results disclosure. Surveys inquired about sociodemographic characteristics, motivations and concerns, behavioral and medical responses to sequencing results, and perceived utility. RESULTS: Among 1395 eligible individuals, 658 enrolled in the Consortium when contacted and 543 have completed a survey after receiving their genomic results thus far (mean age 53.0 years, 61.4% male, 91.7% white, 95.5% college graduates). Most participants (98.1%) were motivated to undergo sequencing because of curiosity about their genetic make-up. The most commonly reported concerns prior to pursuing sequencing included how well the results would predict future risk (59.2%) and the complexity of genetic variant interpretation (56.8%), while 47.8% of participants were concerned about the privacy of their genetic information. Half of participants reported discussing their genomic results with a healthcare provider during a median of 8.0 months after receiving the results; 13.5% reported making an additional appointment with a healthcare provider specifically because of their results. Few participants (< 10%) reported making changes to their diet, exercise habits, or insurance coverage because of their results. Many participants (39.5%) reported learning something new to improve their health that they did not know before. Reporting regret or harm from the decision to undergo sequencing was rare (< 3.0%). CONCLUSIONS: Healthy individuals who underwent predispositional sequencing expressed some concern around privacy prior to pursuing sequencing, but were enthusiastic about their experience and not distressed by their results. While reporting value in their health-related results, few participants reported making medical or lifestyle changes.

Impacts of incorporating personal genome sequencing into graduate genomics education: a longitudinal study over three course years.

BACKGROUND: To address the need for more effective genomics training, beginning in 2012 the Icahn School of Medicine at Mount Sinai has offered a unique laboratory-style graduate genomics course, "Practical Analysis of Your Personal Genome" (PAPG), in which students optionally sequence and analyze their own whole genome. We hypothesized that incorporating personal genome sequencing (PGS) into the course pedagogy could improve educational outcomes by increasing student motivation and engagement. Here we extend our initial study of the pilot PAPG cohort with a report on student attitudes towards genome sequencing, decision-making, psychological wellbeing, genomics knowledge and pedagogical engagement across three course years. METHODS: Students enrolled in the 2013, 2014 and 2015 course years completed questionnaires before (T1) and after (T2) a prerequisite workshop (n = 110) and before (T3) and after (T4) PAPG (n = 66). RESULTS: Students' interest in PGS was high; 56 of 59 eligible students chose to sequence their own genome. Decisional conflict significantly decreased after the prerequisite workshop (T2 vs. T1 p < 0.001). Most, but not all students, reported low levels of decision regret and test-related distress post-course (T4). Each year baseline decisional conflict decreased (p < 0.001) suggesting, that as the course became more established, students increasingly made their decision prior to enrolling in the prerequisite workshop. Students perceived that analyzing their own genome enhanced the genomics pedagogy, with students self-reporting being more persistent and engaged as a result of analyzing their own genome. More than 90% of respondents reported spending additional time outside of course assignments analyzing their genome. CONCLUSIONS: Incorporating personal genome sequencing in graduate medical education may improve student motivation and engagement. However, more data will be needed to quantitatively evaluate whether incorporating PGS is more effective than other educational approaches.

Genetic identification of a common collagen disease in puerto ricans via identity-by-descent mapping in a health system.

Achieving confidence in the causality of a disease locus is a complex task that often requires supporting data from both statistical genetics and clinical genomics. Here we describe a combined approach to identify and characterize a genetic disorder that leverages distantly related patients in a health system and population-scale mapping. We utilize genomic data to uncover components of distant pedigrees, in the absence of recorded pedigree information, in the multi-ethnic BioMe biobank in New York City. By linking to medical records, we discover a locus associated with both elevated genetic relatedness and extreme short stature. We link the gene, COL27A1, with a little-known genetic disease, previously thought to be rare and recessive. We demonstrate that disease manifests in both heterozygotes and homozygotes, indicating a common collagen disorder impacting up to 2% of individuals of Puerto Rican ancestry, leading to a better understanding of the continuum of complex and Mendelian disease.

Concordance between Research Sequencing and Clinical Pharmacogenetic Genotyping in the eMERGE-PGx Study.

There has been extensive debate about both the necessity of orthogonal confirmation of next-generation sequencing (NGS) results in Clinical Laboratory Improvement Amendments-approved laboratories and return of research NGS results to participants enrolled in research studies. In eMERGE-PGx, subjects underwent research NGS using PGRNseq and orthogonal targeted genotyping in clinical laboratories, which prompted a comparison of genotyping results between platforms. Concordance (percentage agreement) was reported for 4077 samples tested across nine combinations of research and clinical laboratories. Retesting was possible on a subset of 1792 samples, and local laboratory directors determined sources of genotype discrepancy. Research NGS and orthogonal clinical genotyping had an overall per sample concordance rate of 0.972 and per variant concordance rate of 0.997. Genotype discrepancies attributed to research NGS were because of sample switching (preanalytical errors), whereas the majority of genotype discrepancies (92.3%) attributed to clinical genotyping were because of allele dropout as a result of rare variants interfering with primer hybridization (analytical errors). These results highlight the analytical quality of clinically significant pharmacogenetic variants derived from NGS and reveal important areas for research and clinical laboratories to address with quality management programs.

Factors associated with success of telaprevir- and boceprevir-based triple therapy for hepatitis C virus infection.

AIM: To evaluate new therapies for hepatitis C virus (HCV), data about real-world outcomes are needed. METHODS: Outcomes of 223 patients with genotype 1 HCV who started telaprevir- or boceprevir-based triple therapy (May 2011-March 2012) at the Mount Sinai Medical Center were analyzed. Human immunodeficiency virus-positive patients and patients who received a liver transplant were excluded. Factors associated with sustained virological response (SVR24) and relapse were analyzed by univariable and multivariable logistic regression as well as classification and regression trees. Fast virological response (FVR) was defined as undetectable HCV RNA at week-4 (telaprevir) or week-8 (boceprevir). RESULTS: The median age was 57 years, 18% were black, 44% had advanced fibrosis/cirrhosis (FIB-4 ≥ 3.25). Only 42% (94/223) of patients achieved SVR24 on an intention-to-treat basis. In a model that included platelets, SVR24 was associated with white race [odds ratio (OR) = 5.92, 95% confidence interval (CI): 2.34-14.96], HCV sub-genotype 1b (OR = 2.81, 95%CI: 1.45-5.44), platelet count (OR = 1.10, per x 104 cells/µL, 95%CI: 1.05-1.16), and IL28B CC genotype (OR = 3.54, 95%CI: 1.19-10.53). Platelet counts > 135 x 103/µL were the strongest predictor of SVR by classification and regression tree. Relapse occurred in 25% (27/104) of patients with an end-of-treatment response and was associated with non-FVR (OR = 4.77, 95%CI: 1.68-13.56), HCV sub-genotype 1a (OR = 5.20; 95%CI: 1.40-18.97), and FIB-4 ≥ 3.25 (OR = 2.77; 95%CI: 1.07-7.22). CONCLUSION: The SVR rate was 42% with telaprevir- or boceprevir-based triple therapy in real-world practice. Low platelets and advanced fibrosis were associated with treatment failure and relapse.

Psychological and behavioural impact of returning personal results from whole-genome sequencing: the HealthSeq project.

Providing ostensibly healthy individuals with personal results from whole-genome sequencing could lead to improved health and well-being via enhanced disease risk prediction, prevention, and diagnosis, but also poses practical and ethical challenges. Understanding how individuals react psychologically and behaviourally will be key in assessing the potential utility of personal whole-genome sequencing. We conducted an exploratory longitudinal cohort study in which quantitative surveys and in-depth qualitative interviews were conducted before and after personal results were returned to individuals who underwent whole-genome sequencing. The participants were offered a range of interpreted results, including Alzheimer's disease, type 2 diabetes, pharmacogenomics, rare disease-associated variants, and ancestry. They were also offered their raw data. Of the 35 participants at baseline, 29 (82.9%) completed the 6-month follow-up. In the quantitative surveys, test-related distress was low, although it was higher at 1-week than 6-month follow-up (Z=2.68, P=0.007). In the 6-month qualitative interviews, most participants felt happy or relieved about their results. A few were concerned, particularly about rare disease-associated variants and Alzheimer's disease results. Two of the 29 participants had sought clinical follow-up as a direct or indirect consequence of rare disease-associated variants results. Several had mentioned their results to their doctors. Some participants felt having their raw data might be medically useful to them in the future. The majority reported positive reactions to having their genomes sequenced, but there were notable exceptions to this. The impact and value of returning personal results from whole-genome sequencing when implemented on a larger scale remains to be seen.

Visualization and cellular hierarchy inference of single-cell data using SPADE.

High-throughput single-cell technologies provide an unprecedented view into cellular heterogeneity, yet they pose new challenges in data analysis and interpretation. In this protocol, we describe the use of Spanning-tree Progression Analysis of Density-normalized Events (SPADE), a density-based algorithm for visualizing single-cell data and enabling cellular hierarchy inference among subpopulations of similar cells. It was initially developed for flow and mass cytometry single-cell data. We describe SPADE's implementation and application using an open-source R package that runs on Mac OS X, Linux and Windows systems. A typical SPADE analysis on a 2.27-GHz processor laptop takes ∼5 min. We demonstrate the applicability of SPADE to single-cell RNA-seq data. We compare SPADE with recently developed single-cell visualization approaches based on the t-distribution stochastic neighborhood embedding (t-SNE) algorithm. We contrast the implementation and outputs of these methods for normal and malignant hematopoietic cells analyzed by mass cytometry and provide recommendations for appropriate use. Finally, we provide an integrative strategy that combines the strengths of t-SNE and SPADE to infer cellular hierarchy from high-dimensional single-cell data.

Development and clinical application of an integrative genomic approach to personalized cancer therapy.

BACKGROUND: Personalized therapy provides the best outcome of cancer care and its implementation in the clinic has been greatly facilitated by recent convergence of enormous progress in basic cancer research, rapid advancement of new tumor profiling technologies, and an expanding compendium of targeted cancer therapeutics. METHODS: We developed a personalized cancer therapy (PCT) program in a clinical setting, using an integrative genomics approach to fully characterize the complexity of each tumor. We carried out whole exome sequencing (WES) and single-nucleotide polymorphism (SNP) microarray genotyping on DNA from tumor and patient-matched normal specimens, as well as RNA sequencing (RNA-Seq) on available frozen specimens, to identify somatic (tumor-specific) mutations, copy number alterations (CNAs), gene expression changes, gene fusions, and also germline variants. To provide high sensitivity in known cancer mutation hotspots, Ion AmpliSeq Cancer Hotspot Panel v2 (CHPv2) was also employed. We integrated the resulting data with cancer knowledge bases and developed a specific workflow for each cancer type to improve interpretation of genomic data. RESULTS: We returned genomics findings to 46 patients and their physicians describing somatic alterations and predicting drug response, toxicity, and prognosis. Mean 17.3 cancer-relevant somatic mutations per patient were identified, 13.3-fold, 6.9-fold, and 4.7-fold more than could have been detected using CHPv2, Oncomine Cancer Panel (OCP), and FoundationOne, respectively. Our approach delineated the underlying genetic drivers at the pathway level and provided meaningful predictions of therapeutic efficacy and toxicity. Actionable alterations were found in 91 % of patients (mean 4.9 per patient, including somatic mutations, copy number alterations, gene expression alterations, and germline variants), a 7.5-fold, 2.0-fold, and 1.9-fold increase over what could have been uncovered by CHPv2, OCP, and FoundationOne, respectively. The findings altered the course of treatment in four cases. CONCLUSIONS: These results show that a comprehensive, integrative genomic approach as outlined above significantly enhanced genomics-based PCT strategies.

Personal Genome Sequencing in Ostensibly Healthy Individuals and the PeopleSeq Consortium.

Thousands of ostensibly healthy individuals have had their exome or genome sequenced, but a much smaller number of these individuals have received any personal genomic results from that sequencing. We term those projects in which ostensibly healthy participants can receive sequencing-derived genetic findings and may also have access to their genomic data as participatory predispositional personal genome sequencing (PPGS). Here we are focused on genome sequencing applied in a pre-symptomatic context and so define PPGS to exclude diagnostic genome sequencing intended to identify the molecular cause of suspected or diagnosed genetic disease. In this report we describe the design of completed and underway PPGS projects, briefly summarize the results reported to date and introduce the PeopleSeq Consortium, a newly formed collaboration of PPGS projects designed to collect much-needed longitudinal outcome data.

Impact of Genomic Counseling on Informed Decision-Making among ostensibly Healthy Individuals Seeking Personal Genome Sequencing: the HealthSeq Project.

Personal genome sequencing is increasingly utilized by healthy individuals for predispositional screening and other applications. However, little is known about the impact of 'genomic counseling' on informed decision-making in this context. Our primary aim was to compare measures of participants' informed decision-making before and after genomic counseling in the HealthSeq project, a longitudinal cohort study of individuals receiving personal results from whole genome sequencing (WGS). Our secondary aims were to assess the impact of the counseling on WGS knowledge and concerns, and to explore participants' satisfaction with the counseling. Questionnaires were administered to participants (n = 35) before and after their pre-test genomic counseling appointment. Informed decision-making was measured using the Decisional Conflict Scale (DCS) and the Satisfaction with Decision Scale (SDS). DCS scores decreased after genomic counseling (mean: 11.34 before vs. 5.94 after; z = -4.34, p < 0.001, r = 0.52), and SDS scores increased (mean: 27.91 vs. 29.06 respectively; z = 2.91, p = 0.004, r = 0.35). Satisfaction with counseling was high (mean (SD) = 26.91 (2.68), on a scale where 6 = low and 30 = high satisfaction). HealthSeq participants felt that their decision regarding receiving personal results from WGS was more informed after genomic counseling. Further research comparing the impact of different genomic counseling models is needed.

Toward clinical genomics in everyday medicine: perspectives and recommendations.

Precision or personalized medicine through clinical genome and exome sequencing has been described by some as a revolution that could transform healthcare delivery, yet it is currently used in only a small fraction of patients, principally for the diagnosis of suspected Mendelian conditions and for targeting cancer treatments. Given the burden of illness in our society, it is of interest to ask how clinical genome and exome sequencing can be constructively integrated more broadly into the routine practice of medicine for the betterment of public health. In November 2014, 46 experts from academia, industry, policy and patient advocacy gathered in a conference sponsored by Illumina, Inc. to discuss this question, share viewpoints and propose recommendations. This perspective summarizes that work and identifies some of the obstacles and opportunities that must be considered in translating advances in genomics more widely into the practice of medicine.

Motivations, concerns and preferences of personal genome sequencing research participants: Baseline findings from the HealthSeq project.

Whole exome/genome sequencing (WES/WGS) is increasingly offered to ostensibly healthy individuals. Understanding the motivations and concerns of research participants seeking out personal WGS and their preferences regarding return-of-results and data sharing will help optimize protocols for WES/WGS. Baseline interviews including both qualitative and quantitative components were conducted with research participants (n=35) in the HealthSeq project, a longitudinal cohort study of individuals receiving personal WGS results. Data sharing preferences were recorded during informed consent. In the qualitative interview component, the dominant motivations that emerged were obtaining personal disease risk information, satisfying curiosity, contributing to research, self-exploration and interest in ancestry, and the dominant concern was the potential psychological impact of the results. In the quantitative component, 57% endorsed concerns about privacy. Most wanted to receive all personal WGS results (94%) and their raw data (89%); a third (37%) consented to having their data shared to the Database of Genotypes and Phenotypes (dbGaP). Early adopters of personal WGS in the HealthSeq project express a variety of health- and non-health-related motivations. Almost all want all available findings, while also expressing concerns about the psychological impact and privacy of their results.

A dynamical model improves reconstruction of handwriting from multichannel electromyographic recordings.

In recent years, several assistive devices have been proposed to reconstruct arm and hand movements from electromyographic (EMG) activity. Although simple to implement and potentially useful to augment many functions, such myoelectric devices still need improvement before they become practical. Here we considered the problem of reconstruction of handwriting from multichannel EMG activity. Previously, linear regression methods (e.g., the Wiener filter) have been utilized for this purpose with some success. To improve reconstruction accuracy, we implemented the Kalman filter, which allows to fuse two information sources: the physical characteristics of handwriting and the activity of the leading hand muscles, registered by the EMG. Applying the Kalman filter, we were able to convert eight channels of EMG activity recorded from the forearm and the hand muscles into smooth reconstructions of handwritten traces. The filter operates in a causal manner and acts as a true predictor utilizing the EMGs from the past only, which makes the approach suitable for real-time operations. Our algorithm is appropriate for clinical neuroprosthetic applications and computer peripherals. Moreover, it is applicable to a broader class of tasks where predictive myoelectric control is needed.

Genetic fine mapping and genomic annotation defines causal mechanisms at type 2 diabetes susceptibility loci.

We performed fine mapping of 39 established type 2 diabetes (T2D) loci in 27,206 cases and 57,574 controls of European ancestry. We identified 49 distinct association signals at these loci, including five mapping in or near KCNQ1. 'Credible sets' of the variants most likely to drive each distinct signal mapped predominantly to noncoding sequence, implying that association with T2D is mediated through gene regulation. Credible set variants were enriched for overlap with FOXA2 chromatin immunoprecipitation binding sites in human islet and liver cells, including at MTNR1B, where fine mapping implicated rs10830963 as driving T2D association. We confirmed that the T2D risk allele for this SNP increases FOXA2-bound enhancer activity in islet- and liver-derived cells. We observed allele-specific differences in NEUROD1 binding in islet-derived cells, consistent with evidence that the T2D risk allele increases islet MTNR1B expression. Our study demonstrates how integration of genetic and genomic information can define molecular mechanisms through which variants underlying association signals exert their effects on disease.