Reproducible Research in Brain Imaging Genetics - Summary With advances in modern technologies, an enormous amount of high-throughput neuroimaging and genomics data are available through large-scale public databases such as the Adolescent Brain Cognitive Development Study (ABCD) project. These expansive big multi-modal data repositories have spurred the rapid growth of brain imaging genomics studies to improve understanding of the pathophysiological processes and genetic architecture of mental disorders. However, these large-scale databases also present significant challenges in data management and analysis, given the common characteristics of large open data including high dimensionality, data inconsistency, and complex correlation structures. In addition, a wide gap exists between neuroimaging and statistical genomics researchers, causing extra barriers to reproducible research in brain imaging genomics. While many training programs provide neuroimaging or genetics data analysis training, very limited (if any) of them focus on studying the relationship between neuroimaging and genomics underlying mental health. This proposal outlines a three-week summer workshop designed to equip graduate-level students or junior researchers with knowledge and skills to conduct reproducible analysis of brain imaging genomics. The primary focus will be on learning in best practices how to estimate meaningful associations of genetic variants on brain structure and function, along with their implications for cognitive functions and broader mental health outcomes. Our proposed workshop will address the critical need for training on neuroimaging genomics to build a bridge between cognitive neuroscience and genomics and promote rigorous and robust analyses of large open brain imaging genomics data with statistical considerations that are specific to large open data . The workshop also aims to promote diversity with trainees having different social and academic backgrounds and prioritizing the attendance of underrepresented scholars. The workshop will feature in-person training with an interdisciplinary faculty team comprising experts in statistical and machine learning methodology, genomics, and brain imaging research. We expect that by end of the workshop the trainees will be able to (1) Learn to access and use large, open brain research databases and understand analytical and statistical considerations specific to large open data; (2) know the important scientific questions in brain research and learn to design and conduct neuroimaging-genotype association studies with clinically meaningful phenotypes; and (3) develop best practices of coding and documentation for reproducible research of brain imaging genomics. All lectures, codes, and lab exercises will be made available on GitHub and the course website, ensuring the widespread dissemination of our training program within the broader community.
