Iuliana Ionita, Ph.D. - US grants

DESCRIPTION (provided by applicant): Common diseases, such as bipolar disorder, asthma, heart disease, cancer, etc. are caused by a complex interplay among multiple genetic and environmental risk factors. Both common and rare genetic variants are expected to influence risk to these traits. Thus far, most research in nding disease susceptibility variants has focused, out of necessity, on the discovery of common susceptibility variants (i.e. variants with a population frequency of at least 5%). Genome-wide association studies have been very successful at nding common variants robustly associated with many complex traits. However, taken together, these variants only explain a small fraction of the estimated trait heritability. Recent advances in sequencing technologies have brought along substantial reductions in cost and in- creases in genomic throughput by more than three orders of magnitude. These developments have lead to an increasing number of sequencing studies being performed, including the 1000 Genomes Project, with the main goal to identify rare genetic variants. Therefore, for the first time, it is now possible to systematically assess the role rare variants may play in various complex traits. Existing methods for the detection of common susceptibility variants are not suitable for the detection of rare variants. We believe that there is a great need for new developments in statistical methodology for the analysis of rare variants, if we want to make the best use of the sequence data currently being generated. The proposed research intends to develop novel and efficient statistical approaches to address this need. The methods proposed here exploit information about the full frequency distributions of rare variants for cases and controls to achieve substantial increases in power over current methods, and can handle large genomic regions, possibly entire genomes. We plan to test our methods on data simulated under a comprehensive set of disease models, and then to apply them to real data on psychiatric diseases, for which common susceptibility variants are very hard to identify. The methods proposed here will be implemented into a software package, to be made available to the larger research community. We believe that this proposal has the strong potential to help in the current efforts to expand the search for causal genetic variants to the, until now, unexplored territory of rare variation. This next phase is key to advancing our understanding of the biological underpinnings of complex diseases, and ultimately essential to improving the public health. PUBLIC HEALTH RELEVANCE: Recent advances in sequencing technologies allow for the first time in history the systematic assessment of the potential role rare variants may play in various complex diseases, such as bipolar disorder and asthma. We propose to develop powerful statistical methods toward this goal, and implement them into a software package, to be made available to the larger research community.

This research is to develop a comprehensive set of statistical methods applicable to high-dimensional and sparse datasets as currently generated by high-throughput genomic experiments. The PI proposes to introduce a novel unified statistical framework to handle large-scale genetic data, and to study its theoretical properties. Within this unified framework, a number of relevant problems can be addressed. The PI will investigate a series of association testing strategies with rare genetic variants that complement and generalize available methods for case-control designs to general designs involving individuals related in an arbitrary fashion. Moreover, the PI will develop an analytic theory to investigate the most powerful statistical designs for association studies with rare variants. The proposed methods will be tested on a broad range of simulated data, and real data from the PIs' collaborators.

The recent progress in genomic technologies has lead to large amounts of genetic data being generated. The emergence of such large-scale, sparse genetic data poses great statistical challenges that require novel and powerful approaches to efficiently extract the information contained in the data. While theoretical, the statistical methods proposed here have the potential to directly contribute to the understanding of the genetic mechanisms underlying complex human traits. To maximize their impact, the proposed methods will be implemented into a software package to be made available to the larger scientific community. Beyond its scientific importance, the project has the potential to contribute to the higher goal of improving the public health. The project also has a strong educational component, and will provide valuable research experience for students and postdoctoral fellows.

DESCRIPTION (provided by applicant): We propose to develop novel statistical methods and software tools for disease association testing with rare variants, with particular application to autism. Although genome-wide association studies have led to the discovery of many common variants reproducibly associated with various complex traits, these variants have small effect sizes and overall explain only a small fraction of the total estimated trait heritability. Recent advances in next-generation sequencing technologies allow for the first time an objective assessment of the importance of rare variants in complex diseases. Over the past few years it has become clear from numerous empirical studies that rare variants are an important contributor to disease risk. This is especially compelling for psychiatric diseases, such as schizophrenia and autism, where common disease susceptibility variants have been more difficult to identify. Traditional association testing strategies that have worked well for common variants have low power for the analysis of rare variants, mostly due to the large number of such variants in any genetic region and their low frequency counts in datasets of realistic sizes. Therefore development of powerful methods for rare variant analysis is greatly needed in order to efficiently extract information from the many sequencing datasets currently being generated. In this application we propose novel methods for both population- and family-based designs to identify rare genetic variants that influence risk to complex diseases, with particular application to autism. In particular, we focus on methods development in the following areas: family-based testing strategies for rare variants, unified testing strategies to efficiently combine family-base and population-based studies, and refinement strategies to identify causal rare variants once an overall association at a gene- or region-level has been established. We will implement the new methods in a comprehensive software package to be made available to the scientific community. Furthermore we will apply these methods to whole-exome data from 1000 autism cases, 1000 matched controls, and 500 autism trios. We believe the proposed research is very timely and has the potential to be of great public health importance through direct application to autism, and more broadly to other complex diseases. PUBLIC HEALTH RELEVANCE: Autism and other psychiatric diseases are major public health problems. The proposed statistical methodology with direct application to autism will help in the identification of genetic variants influencing autism risk, with important implications for public health.

? DESCRIPTION (provided by applicant): We propose to develop and apply state-of-the-art statistical methods to identify clusters of rare disease risk variants within large copy-number variable (CNV) regions previously implicated in autism spectrum disorders (ASD) and schizophrenia (SCZ). Although many large CNV regions have been implicated in risk to psychiatric disorders such as ASD and SCZ, the underlying disease genes in these regions are mostly unknown, because these CNVs are large and contain many genes. Furthermore, these CNVs have not been comprehensively investigated using the large whole-exome sequencing (WES) datasets that have become recently available for ASD and SCZ, with more than 20,000 WES samples combined. We propose to take advantage of these new WES data for ASD and SCZ and propose a systematic investigation of the CNVs implicated in these disorders to identify the underlying disease gene(s) within these CNVs. The problem of identifying rare disease risk variants within these CNVs is of great importance to the field, as rare and large CNVs are the most replicable association so far for these psychiatric disorders. Based on previous work from our group, and taking advantage of some of the largest WES studies for ASD and SCZ, the novel scan statistic approaches we propose to develop promise to help substantially in elucidating the disease genes in these CNV regions. In addition, software implementing these methods will be made publicly available for other researchers interested in pursuing similar work. We believe that the proposed research is very timely and has the potential to be of great public health importance through direct application to autism and schizophrenia, and more broadly to other mental diseases.

? DESCRIPTION (provided by applicant): The tremendous progress in massively parallel sequencing technologies enables investigators to obtain genetic information down to single base resolution on a genome-wide scale in a rapid and cost efficient manner. Despite this progress in data generation, it remains very challenging to analyze and interpret these data. The resulting datasets are high dimensional and very sparse, with millions of genetic variants, the vast majority of which are rare in the population. Identifying which of the many genetic variants in a region of interest are true causal variants is very difficult. Indeed, despite enormous progress in identifying robust associations in genome-wide association studies (GWAS) studies, the underlying causal variants for the vast majority of GWAS loci are unknown. The problem of identifying the underlying causal variants is of fundamental importance for understanding precise biological mechanisms. While experimental functional studies are the gold standard, they are expensive and difficult to implement for a large number of variants. Here we propose to develop state of the art and powerful statistical methods that integrate genome-wide functional annotation data with genetic data on a large number of individuals from whole-exome sequencing and GWAS studies of autism and schizophrenia to help us identify the true causal variants among the abundant natural variation that occurs at a particular locus of interest. The proposed statistical methods will be implemented into a publicly available software package. We believe that the proposed research is very timely and has the potential to be of great public health importance through direct application to autism and schizophrenia, and more broadly to other psychiatric diseases.

Alzheimer?s Disease (AD) is a complex, heterogeneous disorder, and risk to AD is influenced partly by genetics. Understanding the genetic mechanisms that play a role in disease is important as it can lead to a better understanding of the underlying molecular mechanisms, and can identify new gene targets for therapeutic development. We propose gene centric approaches that leverage diverse omics datasets developed specifically for AD (such as AMP-AD), but also more general resources such as GTEx, PsychENCODE, ENCODE, and Roadmap Epigenomics. We will develop quantile tools for transcriptome-wide association studies (TWAS), which are generalizations of TWAS to more complex and heterogenous scenarios where the linear assumptions in standard TWAS are likely to fail. We will also develop gene-based tests using data from WGS by jointly analyzing coding and regulatory variation in predicted regulatory elements likely to affect the expression of a gene under consideration. We will implement these analytical tools into software packages to be made freely available to the community. We will also apply them to some of the largest existing genetic datasets for AD, both GWAS and WGS, and will make the results available to the community on a specially designed web portal.

Summary Labeling clinical data from electronic health records (EHR) in health systems requires extensive knowledge of human expert, is time-consuming, and leads to inconsistencies in case de nitions across di erent phenotyping algorithms. There is increased recognition that common diseases are not discrete entities but rather reside on a continuum. We propose here to take advantage of rich phenotype data in electronic health records, and propose quantitative disease risk scores based on unsupervised methods that require minimal input from clinicians. We will implement the proposed methods into R packages to be made available to the scienti c community. Fur- thermore, we propose applications to phenotypic and genomic data on approximately 100,000 individuals in the eMERGE network, and 500,000 individuals in the UK biobank. We will design a website containing the results of these analyses, including summary statistics from the GWAS analyses for these phenotypes. We believe the proposed research is very timely and novel, and has the potential to facilitate genomic research using rich phenotype data in electronic health records in general.

PROJECT SUMMARY/ABSTRACT The National Academy of Sciences describes the barriers that scientists experience on the path from mentored to independent research as the single most critical threat to the future of biomedical sciences in the United States (US). Women and individuals from marginalized groups face steeper obstacles, particularly in technology-intensive fields such as omic sciences. Despite NIH?s separate grant reviews and paylines, the mean age at which investigators receive their first R01 has continued to climb to an all-time high of 46 years, extending a 35-year negative trend. In parallel, the likelihood of PhD graduates becoming professors has dropped significantly in the last decades, putting the future of biomedical research in the US at risk. To address these challenges, we propose the 'Career MODE' program: Careers through Mentoring and training in Omics and Data for Early-stage investigators to empower diverse cohorts of young researchers across the US?i.e., postdocs in the final years of mentored training and faculty members within two years of their first appointment?to establish independent, successful careers in omics and data sciences. Omics? ranging from genomics to the microbiome?have revolutionized biomedical sciences. As they have become part of any biomedical field, the scarcity of a workforce within each discipline with appropriate omic training has emerged as a critical challenge. We seek to develop an 11-month program to provide intensive training in omics and data science and specialized mentoring from a nationwide, diverse network of 70+ mentors. Career MODE will also feature training in professional skills, including leadership and team management, grant writing, mentoring, goal setting and strategy development, communication and teaching, scientific rigor, transparency, reproducibility, responsible conduct of research, and health equity. By leveraging its nationwide group of mentors and trainees, Career MODE will also offer a structured approach to enhance collaborations and networking, help create teams for breakthrough research projects and K99/R00 and R01-type grant proposals, further career development, and help secure tenure-track positions and faculty promotions. Career MODE will feature hybrid (virtual and in-person) coursework and learning activities (boot camps, roundtables, panels, symposia, observerships), a robust evaluation of each objective, and will generate and disseminate extensive educational material on omics and data science. The program will do outreach to the most promising, motivated, and diverse junior investigators?a group traditionally with limited resources? to create empowered cohorts of scientists to pursue independence through advanced knowledge and tools in omics and data sciences applicable to all biomedical research fields. Thus, our program aligns with NIGMS goals and provides a high return for investment. We will also develop strategies to advance equity and access in omics research and promote Career MODE to academic institutions, sponsors, and stakeholders to reach financial sustainability and self-support beyond the five years of R25 funding.