Michigan Genomics Initiative (MGI)
University of Michigan Biobank[edit]
The Michigan Genomics Initiative (MGI) is a collaborative research effort among physicians and researchers at the University of Michigan with the goal of combining patient electronic medical records with genetic data to gain novel biomedical insights. Surgical patients from the UM Health System are invited to participate, and blood samples are collected and stored in the UM Central Biorepository for future research.
- Genotyping on a customized Illumina HumanCoreExome array (~600K variants)
- HRC imputation (~39M variants)
- EHR-based case-control phenotypes from ICD billing codes International Classification of Diseases (ICD) standard, 9th revision
- Quantitative lab and observational measures
- Anthropometric phenotypes including height and BMI
Project Goals[edit]
Our primary objective is to evaluate if genome-wide analysis of the samples and data being collected through MGI can reproduce and extend the results of previous high quality studies. For example, using the DNA samples we are collecting and EHR-derived phenotypes, can we recapitulate current findings in the areas of diabetes genetics, cardiovascular disease genetics, auto-immune disease, cancer and psychiatric genetics, and other complex diseases? Can we identify new loci for any of these conditions? To address this, we aim to:
- Perform genome-wide PheWAS analysis using ICD codes and imputed genotypes to identify novel associations
- Develop PheWeb browser to display and investigate GWAS/PheWAS results for all ICD code-derived traits and quantitative lab and observational measurements
- Allow researchers to build their own PheWebs to view their GWAS results (e.g. for SardiNIA, Vanderbilt, Partners Biobank, HUNT, etc.)
Results[edit]
- Associations for a total of 1,448 traits with at least 20 cases can be explored in dynamic GWAS, LocusZoom, and PheWAS landscapes via our interactive browser, PheWeb
- We applied a robust and computationally efficient Saddlepoint Approximation Test for PheWAS
Key Individuals[edit]
Leadership[edit]
- Gonçalo Abecasis
- Sachin Kheterpal
- Chad Brummett
- Victoria Blanc
Data Analysis[edit]
- Ellen Schmidt
- Lars Fritsche
- Xutong Zhao
- Joshua Weinstock
PheWeb Architecture[edit]
- Peter VandeHaar
