Leveraging Chemistry and Structural Biology to Predict if a Genetic Mutation Can Cause or Contribute to Disease

Friday, September 25, 2020

Dr. Brett Kroncke

Vanderbilt University Medical Center

Vanderbilt University

Time: 4:45 PM

Location: Hand 1100


The widespread use of genetic testing has led to the association of many variants with disease. However, large phenotyping efforts coupled with DNA sequencing in general populations has exposed a high burden of both rare variants in disease-associated genes and previously annotated pathogenic variants. In this talk I will describe how we tested the hypothesis that the presence of one of these variants confers a knowable probability of disease equivalent to the positive predictive value of that variant. To test this hypothesis, we chose to evaluate variants in the well-characterized cardiac potassium channel gene, KCNH2, and their association with the clinically well-defined long QT syndrome. From and extensive literature review, we estimated the probability of long QT diagnosis for each KCNH2 variant using a Bayesian method that integrates together patient data and variant features (changes in variant function, protein structure, and in silico predictions). We validated these estimates on an international cohort (France, Italy, and Japan) of 933 individuals heterozygous for 266 missense or in-frame insertion/deletion variants in KCNH2. We suggest our framework can translate variant location in three-dimensional space, in vitro functional data, and in silico predictors into the information equivalent of 10 heterozygous individuals for each variant. Most importantly, these data can be obtained in the absence of any clinical observations of heterozygotes.


For my graduate work at the University of Virginia, I focused on experimental methods development for determining membrane protein flexibility and membrane protein structure. As a Postdoctoral Fellow at Vanderbilt, I further pursued translational opportunities of experimental and computational structural biology. The last two years of my training focused on learning from experts in the fields of cardiac arrhythmia genetics, and computational phenotype-predictive modeling to construct predictive models of ion channel phenotypes and validate the resulting predictions. My long-term research objective is to determine what genetic variants mean to the individuals who carry them. A reasonable and tractable starting place to develop and prove these new methodologies is in ion channels responsible for heart arrhythmias, specifically the three genes most associated with Long QT syndrome (LQT1-3; KCNQ1, KCNH2, and SCN5A). Given the high estimated frequency of rare mutation carriers in LQT1-3 genes, nearly 2% of the general population, the driving force of this project is not only to provide a reliable estimate of if an individual carrying a variant has high probability to present with disease, but also if that individual has a negligible probability of presenting with disease.

Click here to view more Seminars