Our group develops computational methods to investigate metabolic alterations related to carcinogenesis, drug resistance, and cancer patient outcomes. The ultimate goal is to establish metabolic signatures that can predict response to certain drugs and to rationally engineer effective combination therapies for individual patients. Some of our recent work in this context is highlighted below.
In a pharmacometabolomics project, we investigated the effects of the histone deacetylase inhibitor ‘Panobinostat’ in refractory diffuse large B cell lymphoma (DLBCL) patients, where metabolomic and transcriptomic analysis revealed an upregulation of the choline pathway after treatment. Inhibition of this pathway in vitro and in vivo models led to synergist effects on cell viability, effectively killing cancer cells. This illustrates that metabolic alterations can be leveraged to infer cellular escape mechanisms that can be targeted with therapeutic interventions.
To further explore the potential of targeted metabolic interventions, we have initiated a project aimed at leveraging the metabolic alterations of chemotherapy-resistant tumors to rationally design patient-specific combination therapies that will improve response to chemotherapy. We will use an in silico computational network of metabolic and gene regulatory interactions to infer the most promising drug targets, which will then be validated experimentally in patient-derived organoid models.
In collaboration with the Cornell Veterinary Biobank, we are investigating the possibility of using dogs as a spontaneous model for human lymphoma. We have metabolically profiled a cohort of dogs with DLBCL and matched controls, and are currently analyzing these data to identify metabolic markers of disease in canine subjects. The results will then be compared to the metabolic alterations observed in human patients to establish and characterize the similarities between the two species.
Alzheimer’s disease (AD) is a neurodegenerative disease affecting 1 in 9 people over the age of 65. AD is usually diagnosed after the onset of noticeable neuropsychiatric deterioration, such as memory loss, speech impairment, and social withdrawal. Early changes in the brain are suspected to begin over 20 years before any detectable symptoms of the disease. Our lab focuses on two pertinent challenges in the field: Understanding the molecular alterations underlying the disease in order to aid the development of novel therapies and developing early detection approaches that would allow for timely interventions.
Our research is funded by NIH/NIA grants and integrated into several national and international consortia, including the Alzheimer’s Disease Metabolomics Consortium (ADMC) and the Alzheimer’s Gut Microbiome Project (AGMP). We analyze omics data from large cohort studies, including ADNI, ROS/MAP, Mayo Clinic, UK Biobank, and others.
Recently, we published a landscape of metabolic alterations in AD, based on 500 human post-mortem brain samples taken from the ROS/MAP cohort. Currently, we are exploring AD-associated metabolic alterations across four different brain regions, which will provide a more comprehensive picture of AD’s metabolic imprint in the brain. These brain-based association analyses provide novel metabolic insights into AD, inform follow-up experiments in model systems, and potentially pave the way for novel therapeutic strategies.
In other projects, we are investigating personalized dietary or lifestyle interventions that can increase individual resilience against bioenergetic disturbances in AD, studying the molecular crosstalk across the blood-brain barrier, determining the role of the gut microbiome in AD, and developing novel network-based approaches for drug repositioning.
Our lab works on a variety of further research topics, including:
Type 1 and Type 2 diabetes research
Biomarkers and mechanisms of COVID-19 severity
Inference of protein glycosylation pathways from large-scale glycomics data
Novel approaches for network inference based on Gaussian graphical models (GGMs)
Statistical approaches for the preprocessing and analysis of omics data, and metabolomics in particular
Multi-omics integration methods
Variational Autoencoders & Deep Learning