Projects in Mendelian Randomization
Overview
Causal inference is one of the most challenging problems in biology and medicine, requiring multiple lines of evidence supported by model systems, pharmacology, and genetics to demonstrate convincingly. Under specific assumptions, it is possible to use human genetic variation as an instrumental variable to estimate the causal effect of an exposure on an outcome. For example, loss-of-function variation in PCSK9 lowers LDL cholesterol (LDL-C) and susceptibility to heart disease, consistent with the fact that elevated LDL-C is a causal risk factor for heart attack. There is much nuance here, with advances in new methods, understanding the testable (and untestable) assumptions of the model, kits of tools that make application very straight forward – a complete didact on this subject is well beyond the purpose of this short overview. For those who are interested, I strongly recommend picking up a recently published MR Primer:
Mendelian Randomization Analyses
We are generally interested in novel relationship discovery as well as clarifying previously observed exposure-outcome relationships, particularly (but not exclusive to) the cardiometabolic trait space. For example, we have explored the role of smoking and body mass index (BMI) on T2D risk, serum urate on CAD/T2D, and BMI and Waist-Hip ratio (WHR) on blood traits. Ongoing / active projects include:
-
Nutrition / dietary exposure on cardiometabolic traits/disease. Dietary preference is conventionally thought of as an environmental risk factor, but previous studies have elucidated a significant (but small) heritable component to this collection of traits. This facilitates MR experiments to understand what, if any, role this ‘environmental’ exposure has on T2D, CVD, or related causal traits. A leading hypothesis is that risk from dietary preference to outcome is mediated through anthropometric factors like adiposity. This work is led by Mia Lee (PGG graduate student) in the lab.
-
Phenome-wide MR; considerations for diverse ancestry analysis. In collaboration with Drs. Venexia Walker and Tom Gaunt (Univ. of Bristol, UK) and Dr. Scott Damrauer (Penn), we are engaged in a broad stroke, all-by-all MR experiment (i.e., hundreds of exposure-outcome pairings) via summary data from MVP and the UKBB. This will provide a deep, well powered resource for discovery, but will explore the role and extent of biases that must be contended with in the case of ancestry mismatching.
-
Collaborative activities with T2D-GGI. Collaborating within the T2D-GGI, we are engaged in the latest GWAS for T2D and QTL data through which we can ‘drug target’ via QTLs. The goal with this approach is to identify potential leads and directions of effect for changes in molecular phenotype (e.g., expression) with a target gene, to make an educated guess at the predicted effect that would be therapeutic. In addition, work is underway which seeks to relate our previously defined genetic clusters to T2D comorbidities (e.g., retinopathy, stroke, etc.) through the MR framework.
Generally, we are open to different types of analysis projects that are along the lines of specific targeted questions!
Mendelian Randomization Methodology
We are also interested in developing methods in MR studies. Previously, we have developed pipelines and a simulation engine to evaluate (empirically) MR under different models and assumptions.
Methods development for MR is a particular challenge, broadly because the field moves rapidly and there are a great many biostatistician/epidemiologists who have dedicated their entire operation to MR (and methods development). Where possible, we seek to bring in these latest methods into ongoing work. Occasionally, we identify methodologic “targets of opportunity” while working with Big Data sets or by pushing the envelope of human genomics studies. We have a couple of areas of interest in this space, primarily along the lines of instrument curation, and when considering large-scale data sets which derive from a plurality of ancestries. That said, my group is certainly open to other lines of inquiry, so interested parties should express their interest so that we can discuss further.