Translational Cardiogenetics · Vanderbilt

What does a genetic variant mean for the person who carries it?

We complement variant classification with quantitative penetrance estimates: a calibrated probability of disease for each variant, built from published carriers, high-throughput functional data, protein structure and Bayesian modeling.

Beyond the label

Carriers of the same variant can have different outcomes. We quantify that variation.

S1787N · 1 of 13
Y1795H · 5 of 12
R367H · 16 of 19
0% · lower probabilityhigher probability · 100%

Observed proportions of published SCN5A heterozygotes diagnosed with Brugada syndrome: 1/13 (8%), 5/12 (42%), and 16/19 (84%). These historical carrier proportions illustrate variable expression; they are not population prevalence or individual risk predictions. The Bayesian model combines carrier observations with variant features and reports credible intervals. Method: Kroncke et al., PLOS Genet 2020; carrier study: Kroncke et al., Circ Genom Precis Med 2018.

Approach

Three complementary research approaches.

01

Bayesian penetrance

A patented framework that fuses carrier counts, functional data, structural context and in silico predictors into a variant-level probability of disease with explicit uncertainty.

02

Functional genomics

Deep mutational scanning, calibrated automated patch clamp and CRISPR-edited iPSC-cardiomyocytes measure what variants do, at the scale sequencing demands.

03

Auditable AI curation

Large language models extract per-variant carriers and phenotypes from PubMed; a multi-model consensus scores gene–disease validity against the ClinGen SOP, with every call traceable to its source.

Two questions, kept separate

A penetrance estimate describes a variant. Clinical risk describes a person.

Our variant-level estimates give the probability that a heterozygote, on average, is affected. An individual’s risk also depends on polygenic background, sex, QTc, age and treatment. The second question is where the program is heading: patient-derived iPSC lines at the extremes of QT polygenic score, survival models of breakthrough events on therapy that we are now evaluating, and the NHGRI-funded effort to predict clinically important outcomes for people with pathogenic variants.

Disease-associated variants in Nav1.5 mapped onto structure
Variants in NaV1.5 (SCN5A) mapped onto channel structure. Brugada, LQT3 and unaffected carriers cluster differently across the protein.
The public resource

VariantBrowser.org

Free to clinicians, genetic counselors and researchers. Each gene page shows carrier counts, functional data, structural context and predictors alongside the Bayesian estimate, so the evidence behind every number can be inspected.

KCNQ1

Long QT type 1

4,386 variants

KCNH2

Long QT type 2

7,995 variants

SCN5A

Brugada / LQT3

13,739 variants

RYR2

CPVT

29,236 variants

Recent publications

Selected work, 2024–2026

Full list
Who does this work

People

Wet lab, computation and clinic, led by Brett Kroncke, Ph.D.

Meet the lab →
Tools & data

Resources

Open-source curation, feature aggregation and penetrance code on GitHub.

Browse the toolkit →