(Look HERE for rotation projects!)

Our lab is focused on understanding how interactions between different biological levels result in phenotypic change.  We study the associations between DNA variation, DNA methylation, gene expression, environmental factors, and clinical traits to gain a better understanding of how individual differences lead to dramatically different disease outcomes.  We use a combination of bench and bioinformatic techniques to do so, with the ultimate goal of allowing our computational approaches to inform our experimental work, and our experimental work to inform our algorithm development.



Our work utilizes mouse genetic resource populations to study the progression of heart failure, a disease that is notoriously difficult to study using Systems Genetics approaches in human populations.  Our primary resource population is the Hybrid Mouse Diversity Panel (HMDP), which contains over 150 fully inbred strains of mice that show significant genetic and phenotypic diversity and have been used in labs around the world to study diseases ranging from atherosclerosis to bone density to hearing loss.

We combine our biological data with computational approaches, some from other labs, some home-grown, to understand how all of our high-throughput data combine together to create the clinical phenotypes we observe.


Current Projects

1)  Epigenetic Drivers of Heart Failure

Heart failure is a growing concern among researchers, with rates expected to increase by 25% by 2030.  Heart failure is an incredibly complex disease with many pathological features including cardiomyocyte hypertrophy, contractile dysfunction, and fibrotic remodeling.  Heart failure has complex etiologies, with common risk factors such as hypertension and diabetes being in and of themselves multi-factorial.  Consequently, there is a tremendous amount of heterogeneity in human populations in both disease onset and progression which mask the demonstrated strong genetic component of common forms of heart failure.  As a result of this heterogeneity, human genome-wide association studies have only been able to recover a handful of significant loci.  Recently, the PI’s group described a population of mice in which over 30 loci for heart failure-related phenotypes were identified and which demonstrated significant overlap (50%) with significant or suggestive heart failure-associated loci in humans.

Our lab has recently expanded the study of these mice from the genomic scale to the epigenomic scale by performing and analyzing the results of Reduced Representational Bisulfite Sequencing across 92 strains of the Hybrid Mouse Diversity Panel both before and after catecholamine stimulation.  We have identified a number of associations between the methylome and phenotypic traits in both the baseline and catecholamine-treated cohorts.  Our research continues in identifying additional loci of interest and beginning the process of validation and mechanistic discovery using in vitro and in vivo models.

2)  Genetic Drivers of Cellular Composition in the Heart in Healthy and Failing States

The underlying cellular composition of the heart is surprisingly poorly understood.  Different groups, using different techniques and different model systems have reported wildly different proportions of cardiomyocytes to non-cardiomyocytes as well as different compositions of the non-cardiomyocyte component of the heart.  Missing in these analyses is a study of how genetic backgrounds can lead to differences in cellular composition of heart tissue.  By leveraging the structure of the HMDP and performing RNAseq and RRBS at a single-cell level, we aim to obtain estimates for cell-type composition in the heart across the entire HMDP and identify candidate genes and loci which lead to these changes.

3)  Multi-level Co-regulation Networks

In prior work, we developed an approach, weighted Maximal Information Component Analysis (wMICA) to explore transcriptomic co-expression networks in a way that allows for proportional membership within co-expression models and accounts for the non-linear relationships which are commonly observed between gene and protein abundances.  This method has been used to identify novel regulators of cardiac phenotypes and as a hypothesis-generating tool for downstream validation.

The underlying framework of wMICA was designed to be able to incorporate data from any biological layer, however, there remains a number of open questions regarding how to connect data of different types to one another and form more complete hypotheses of the flow of information within healthy and diseased individuals.  We are working on solving these problems as well as further developing the wMICA algorithm to improve robustness and comprehensibility.