Calendar of Relevant Events at Carnegie Mellon University
Title: Generative Method to Discover Genetically Driven Image Biomarkers
Abstract:
We present a generative probabilistic approach to discovery of disease subtypes determined by the genetic variants. In many diseases, multiple types of pathology may present simultaneously in a patient, making quantification of the disease challenging. Our method seeks common co-occurring image and genetic patterns in a population as a way to model these two different data types jointly. We assume that each patient is a mixture of multiple disease subtypes and use the joint generative model of image and genetic markers to identify disease subtypes guided by known genetic influences. Our model is based on a variant of the so-called topic models that uncover the latent structure in a collection of data. We derive an efficient variational inference algorithm to extract patterns of co-occurrence and to quantify the presence of heterogeneous disease processes in each patient. We evaluate the method on simulated data and illustrate its use in the context of Chronic Obstructive Pulmonary Disease (COPD) to characterize the relationship between image and genetic signatures of COPD subtypes in a large patient cohort.
Statistical Inference of Reticulate Evolutionary Histories Using Data from Unlinked Loci
Luay Nakhleh
Department of Computer Science
Rice University
Luay Nakhleh
Using genome-wide data for phylogenetic inference and analysis has become commonplace in the post-genomic era, giving rise to the field of phylogenomics. The multispecies coalescent (MSC) model has emerged as the main stochastic process that helps capture the intricate relationship between species trees and gene trees. Combined with models of sequence evolution, the MSC can be viewed as a generative model of genomic sequence data in the context of a (species) phylogenetic tree.
A significant outcome of the use of genome-wide data has been the increasing evidence, or hypotheses, of reticulation (e.g., hybridization) during the evolution of various groups of eukaryotic species. Reticulate evolutionary histories are best represented as phylogenetic networks, which extend the tree model to allow for admixtures of genetic material. In this talk, I will describe the multispecies network coalescent (MSNC) model, which extends the MSC model so that it operates within the branches of a phylogenetic network. This extended model naturally allows for modeling vertical and horizontal evolutionary processes acting within and across species boundaries. In particular, it simultaneously accounts for gene tree incongruence across loci due to both hybridization and incomplete lineage sorting. I will then describe a likelihood function for this model, as well as a method for Bayesian sampling of phylogenetic networks and their parameters using reversible-jump Markov chain Monte Carlo (RJMCMC). All the methods I describe have been implemented in our open-source software package, PhyloNet, which is publicly available at http://bioinfo.cs.rice.edu/phylonet.
Bio:
Luay Nakhleh is Professor and Chair of the Computer Science Department at Rice University. He received the B.Sc. degree in Computer Science from the Technion (Israel), the Master’s degree in Computer Science from Texas A&M University, and the Ph.D. degree in Computer Science from The University of Texas in Austin. He conducts research in the areas of bioinformatics and computational biology, focusing mainly on questions in evolutionary biology. Luay is a recipient of the DoE CAREER award, the NSF CAREER award, the Sloan Fellowship, and the Guggenheim Fellowship.
Sushmita Roy’s research focuses on developing statistical computational methods to identify the networks driving cellular functions by integrating different types of genome-wide datasets, that measure different aspects of the cellular state.
Roy is interested in identifying networks under different environmental, developmental and evolutionary contexts, comparing these networks across contexts, and constructing predictive models from these networks.
Specifically, some research topics of interest are:
Inference of structure and function of regulatory networks
Comparative analysis of expression modules across species
Evolution of gene regulation
Relational learning to predict function
Modeling condition-specific functional behavior
Learning causal networks
Predictive models of phenotypic response
