Projects

Network modeling for identifying drug targets and disease progression

Network inference tools aim to identify the most relevant dependencies that exist between any two elements given a large set of data. In particular, we explore the relationship between transcriptional regulatory genes and the biological pathways they regulate. Genes that control dysregulated pathways in disease become potential targets for therapeutic intervention. We develop tools that effectively identify these target genes. Additionally, we focus on bottom-up network modeling of gene-gene interactions through Boolean models. The models can be computationally intensive, so we aim to develop computationally efficient models that can be used to simulate the effect of different genetic aberrations and drug interventions.

Characterizing drivers of aggressive prostate cancer

A central challenge in prostate cancer research and clinical treatment is the characterization of the molecular mechanisms that drive aggressive, lethal disease. Roughly 90% of men diagnosed with prostate cancer will die with the disease rather than from the disease. These 90% of men often suffer from adverse and unnecessary side effects of over-treatment. Thus, we focus on developing computational methods to discover the molecular drivers of aggressive disease and validate these drivers using mouse models through collaborative with the Cramer lab. We have already characterized a novel, aggressive molecular subtype and aim to characterize more.

Predicting drug sensitivity from –omics datasets

A central goal of personalized medicine is to determine treatment for patients given their genomic backgrounds. We are far from systematically accomplishing this goal with patients, thus we use human cell lines as a proxy to study the interactions between cells and drugs. We focus on understanding the genomic factors that contribute to drug response or resistance.

Bladder cancer genomics and pharmacology

75,000 new cases of bladder cancer will be diagnosed each year. Little advance has been made in chemotherapeutic treatment of late stage or metastatic bladder cancer. We leverage large amounts of -omics datasets being generate to identify genes that may play a role in tumor progression and potentially affect drug treatment. We run shRNA screens in the presence of standard chemotherapies to find novel drug targets and use network inference methods to model the pathways and processes underlying synergistic drug sensitivity.

DREAM (Dialogue for Reverse Engineering Assessments and Methods) Challenges

In additional to these primary research areas, we are active in the organization of DREAM challenges. The DREAM project is organized around a community of data scientists, where high-impact data along with challenges are presented annually, participants submit their best solutions, and assessment is performed using standardized metrics and blinded gold standards. The product of this effort is a rigorous assessment and performance ranking of the current best methods to address a challenge, along with cultivating a community of scientists interested in biomedical research problems.