Setia-Verma Lab
Integrative omics @ University of Pennsylvania
Setia-Verma Lab
Integrative omics @ University of Pennsylvania
Overview
Research Overview
Genetic, biological, and environmental factors all play a crucial role in influencing common and complex diseases. Thus, it is necessary to align the biological mechanisms with clinical phenotypes while accounting for environmental factors to understand disease etiology. In my view, the alignment of multi-modal biomedical data is necessary to enhance our understanding of the disease architecture. Next-generation health care is moving towards a paradigm of using data-driven models for profiling patients disease risk. My research lab focus on approaches for discovery and implementation science to identify novel indicators of human health and diseases. Our long term goal is to use multi-modal risk prediction models to estimate disease risk and long term effects of a disease.
Electronic Health Record (EHR) Data
Using EHR data for preventive care facilitated by integrating a patient's medical and genomic information is becoming immensely useful in illuminating the underlying architecture of common complex disease. Our lab has a track record in using de-identified EHR and biorepositories for association studies. Association studies include both rare and common variants. We apply methods to characterize variants that impact the function of the genome and implicate the disease etiology.
Computational methods for integrative omics
We focus on the development of a novel framework for data integration that provides meaningful interpretations of statistical models. Previously, we have incorporated relationships between multi-omic variables as an integrative approach to identify models influencing the risk of cancer using a neural network approach. One way to approach data integration in understanding complex heterogeneous disease traits is first to extract clinically homogenous samples and then identify their genetic similarities. We can do this by collecting multi-omic datasets and develop integration methods to identify disease subtypes. These approaches could help explain how genomic features might impact clinical outcomes.
Women's Health Research
Although the number of association studies using EHR-derived phenotypes (both structured and unstructured data) has increased, understanding women's health by applying EHR data remains underutilized. Most phenotypes obtained from diagnostic code-based algorithms are often presented in the absence of longitudinal data even though the majority of women choose their primary care for the pregnancy and postpartum interval after an initial antenatal visit. Thus, the EHR represents a more transparent and concise picture of women's health during this time than many prospective cohort studies, particularly since enrollment of women in research studies can start as early as during their regular visits to obstetrics and gynecology.
. These EHR variables can help us design algorithms to identify true cases and controls for multiple diseases and subgroups of diseases. Linking patient's EHR data to genomic, metabolomic, and other exposome data can also help gain a better understanding of the pathophysiology of women's health.
Precision Medicine
The fundamental goal of precision medicine is to provide the right care, to the right patient, at the right time. My research strategy considers every disease as a new hypothesis and identifies significant features from all available 'omic and clinical phenotypes to be used in modeling disease traits. I believe that with the ability to explore the different types of 'omic variations, by assuming different underlying genetic architecture and constructing data-driven models, we could achieve our maximal understanding of the genomics of common diseases. My lab continues to focus on the drive towards implementing detrimental genetic findings in clinical practice to improve patient outcomes.