BioVU is the Vanderbilt University Medical Center (VUMC) biobank of DNA extracted from discarded blood collected during routine clinical testing and linked to de-identified electronic medical records (EMRs). While DNA has been collected on 270,000 patients, EMRs for over 2.6 million patients are also available for phenotypic studies. I will describe results from three recent biobank initiatives including phenotypic, genetic, and ethical, legal and social issues (ELSI) research.Phenotyping initiative: In collaboration with the Tourette and Obsessive-Compulsive Disorder (OCD) Workgroup of the Psychiatric Genomics Consortium (PGC), we developed a phenotyping algorithm for the detection of OCD in the EMR. We utilized ICD 9 and 10 codes along with natural language processing (NLP) to detect diagnostic keywords (e.g., OCD, etc.), treatment keywords (e.g., cognitive-behavioral, etc.), and medications (i.e., venlafaxine, etc.) commonly prescribed for the treatment of OCD. Contextual information from 50 cases identified by the algorithm was then abstracted by trained non-expert reviewers and abstracted charts were reviewed by two domain experts to determine the positive predictive value (PPV) of the algorithm. We identified 7,547 potential cases in the EMR and PPV was high (> 92%) across the two reviewers when ICD codes, diagnostic keywords, and treatment keywords were present in case charts. I will present our algorithm development approach, pitfalls and challenges, validation measures, and patterns of comorbidity discovered within the EMR.Polygenic analysis initiative: We have developed an EMR-enabled biomarker discovery approach which leverages routinely collected clinical laboratory measurements available on >30,000 people with genetic data in BioVU. These lab measures such as complete blood counts or metabolic and immunoglobulin panels can be mined for novel biomarkers through polygenic risk score (PRS) analyses. In an initial experiment, we calculated PRS for triglycerides (PRStrigs) and coronary artery disease (PRSCAD) and demonstrate that each predictor, respectively, accounted for a significant proportion of the variance in mean triglyceride levels across the biobank (PRStrigs; R2 = 0.034; p = 4.2x10-27 and PRSCAD; R2 = 0.0051, p = 2.2x10-5) while adjusting for sex, age, and ancestry. In a separate analysis, we selected 53 putative biomarkers measured in the Atherosclerosis Risk in Communities (ARIC) study (n=7,740 unrelated subjects) and used Bayesian sparse linear mixed modeling to calculate genetic scores across 37,153 individuals with EMR data available. We discovered that a genetic predictor for white blood cell count was associated with anxiety disorder (OR=1.07, [1.04-1.10]) in this sample. These and other novel results from ongoing analyses will be presented.ELSI Initiative: Biobank research poses unique ethical challenges. The Center for Genetic Privacy and Identity in Community Settings (GetPreCiSe) at VUMC is an NIH Center of Excellence in ELSI Research (CEER) aimed at a) developing a more complete understanding of privacy concerns related to data sharing and b) addressing these concerns through policy recommendations. Early studies from the Precision Medicine Institute (PMI) and GetPreCiSe indicate that there is widespread confusion and concern about the meaning of “genomic privacy” in the context of biobanks. I will share the results of these early studies and discuss how they are shaping biobank research at VUMC and PMI.