|
"Humans are exposed to a wide range of drugs over the course of their lives; sometimes decades of exposure of drugs for chronic diseases. While the basic safety and efficacy in treating the target disease have been well-established for all common drugs, we know little about the wider but subtle effects those drugs may have on other aspects of health. Arguably the most important such aspect is lifespan: understanding how drugs affect lifespan could reveal new mechanisms and potential inhibitors of aging. Since most common drugs are administered orally and absorbed in the gut, one of the first interactions drugs have with the host is via its gut microbiome. Several recent studies have shown that ~90% of common drugs do in fact perturb the gut microbiome, which is well-established to play a critical role in human health. Together these points raise the hypothesis that the gut microbiome may mediate the effects of drugs on lifespan.
My collaborator, Dr. Alexey Ryazanov (Department of Pharmacology, RWJMS), has performed the largest ever screen of various drugs on mouse lifespan. He tested more than 1000 drugs belonging to 62 different pharmacological classes that span the entire pharmacological space with ~75% of drugs currently used in medicine. The primary endpoint was mouse mortality, from which he calculated median and maximal lifespan for each experimental group. During the study he collected > 7,000 necropsies from mice administered all tested drugs throughout lifespan. Now with additional collaborators Dr. Rohan Maddamsetti (Department of Biochemistry and Microbiology, RU-NB) and Dr. Maria Gloria Dominguez-Bello (Department of Biochemistry and Microbiology, RU-NB), we will build on this preliminary data to test the hypothesis that the gut microbiome is a key mediator of the effects of drugs on lifespan.
The aim of this undergraduate research project is to build a comprehensive AI-ready database linking drug screens to microbiome sequencing data in humans and mouse models. We will combine all existing large-scale drug screens with our data on lifespan changes and gut microbiome perturbations. We will use this as a resource for training machine learning (ML) models to predict drug effects. Specifically, we will train a white-box ML model in which microbiome composition is an explicit intermediate layer in a deep neural network that takes a drug as input and ultimately predicts an effect on lifespan. This resource will be a foundation for future AI training endeavors and will let us estimate the effect size of microbiome changes on lifespan.
This work will be the first comprehensive analysis testing for a correspondence between long-term drug administration and microbiome composition in a mouse model. This analysis will lead to a 1) better fundamental understanding of microbiome mechanisms affecting health and lifespan and 2) predictive statistical models for how drug treatment perturb the gut microbiome and quantitatively affect lifespan outcomes."
|
Sign in
to view more information about this project.
