Current studies of DNA repair processes are focused on dissecting the function of single genes. This classic molecular biology approach provides a wealth of knowledge about individual components and their functions in DNA repair; however, this approach is not able to integrate all information about properties of individual DNA repair genes or proteins to elucidate their collective interplay. To overcome this challenge, we proposed to study DNA repair using systems biology approaches. Homologous recombination (HR)-mediated DNA repair is a fundamental biological process to prevent genomic instability by repairing DNA double strand breaks (DSBs). We have generated a HR defective (HRD) gene signature via genome-wide transcriptome profiling in HR-deficient cells. Our study showed that the HRD signature can be used to functionally assess HR repair status at a network level without interrogating specific genetic alterations in cells. HRD can also serve as an effective drug discovery platform to identify drugs targeting HR repair as potential chemosensitizers. More importantly, this HRD is able to predict clinical outcomes across multiple tumor lineages by functionally assessing HR repair status in patient tumor samples. Furthermore, using data in the Ingenuity databases, we constructed a graph with nodes representing genes and directed links representing biological interactions. We were able to analyze the HR repair network as a weighted directed graph, to which we used flow network and maximum flow to predict mechanisms and/or targets that can synthetically rescue the dysfunctional HR repair network in BRCA-deficient cells. It is our hope that the knowledge gained from our study may help develop new strategies for preventing BRCA-associated cancer in women to realize the promise of precision medicine in cancer prevention.
Dr. Guang Peng is an assistant professor in the Department of Clinical Cancer Prevention at The University of Texas MD Anderson Cancer Center. The long-term goal of her research is to characterize and target molecular regulators of the mutational and dysfunctional DNA repair processes driving tumorigenesis. One of her major interests is to develop systems biology approaches to understand the dynamics of the DNA repair network in tumor evolution. We aim to utilize network-based mathematical modeling and molecular biology approaches to understand and target the DNA repair network. These new interdisciplinary approaches will offer a revolutionary conceptual framework to determine the compound effect of the DNA repair network rather than focusing on an individual repair gene’s function in tumorigenesis. Her research has been funded by several agents including National Cancer Institute (NIH), Department of Defense, American Association for Cancer Research, Susan Komen Foundation and Cancer Prevention Research Institute of Texas(CPRIT) .
Date, Time & Location: Monday, April 18th at 3pm in MP 3314
Speaker: Dr. K.T. Arasu, Wright State University
Title: Sequences Having Good Correlation Properties With Applications To Radar And Communication Systems
Abstract: In this day and age of high tech and big data, our modern lives literally revolve around sequences, as data streams are sequences of varying lengths. Binary sequences, along with their multi-phase analogs, whose periodic and/or aperiodic correlations possess desirable properties are now applied in our day-to-day lives, thanks to a plethora of research area : coding theory and cryptography, radar distance ranging, synchronization, identification, and hardware testing, error correction codes, CDMA standard for cell phone communications, to name a few.
In this talk, we provide (i) a quick overview of how mathematical structures from algebra, number theory, finite geometry and discrete mathematics have important ramifications in applications areas of information engineering and (ii) the mathematical ideas used in some recent construction techniques of certain new families of sequences and arrays with desirable correlation properties.
Date, Time & Location: Friday, April 8, 2016 at 2:30 pm – 3:20 pm in MP 3314
Speaker: Joint Computational Sciences, Statistics and Disease Dynamics Seminar, Dr. Alexandra Smirnova, Georgia State University
Title: On Stable Estimation of System Parameters in Infectious Disease Modeling.
Abstract: Parameter estimation problems in epidemiology and infectious diseases constitute a large class of unstable (ill-posed) models. These problems come with some unique challenges. (1) In the past years, models developed with annualized data primarily utilized constant system parameters (SP). However, with the advent of more timely and frequent reporting of clinical data, some SP emerge as time dependent. With variable SP, the dimensionality of the solution space is growing, and the optimization problem becomes under-determined. In order to solve it in a stable fashion, a rather sophisticated “problem-oriented” regularizing algorithm must be proposed. (2) One of the main dangers of instability is that we can get a very good fit, but with highly inaccurate SP. Thus, regularization is essential, but the noise level in clinical data is almost impossible to estimate. For this reason, one has to resort to heuristic methods for the evaluation of a regularization parameter (RP). Oftentimes, these methods provide us with some insights into how RP can be selected, but the choice is not theoretically justified. (3) The inherently differing scales of biological parameters in disease models may complicate their simultaneous recovery by a regularized optimization algorithm based on the original Tikhonov functional with L2 stabilizer. Specially designed penalty terms, capable of balancing the sensitivity of the cost functional to multiple system parameters, need to be constructed for linear and nonlinear optimization algorithms. In my talk, the question of possible systematic errors in heuristic methods and other important aspects of practical implementation of data-driven algorithms for the selection of a regularization parameter, combined with different stabilizing procedures, will be investigated and tested on synthetic and real data for inverse problems in epidemiology. This is a joint work with G. Chowell, M. Marcheva, N. Tuncer, A. Akossi, L. deCamp, M. Sheppard.
Revisiting Mathematical Models of the Ebola Epidemic in West Africa: Principles, Predictions and Control
Mathematical models of disease dynamics were utilized during the Ebola virus disease (EVD) epidemic in W. Africa to predict the potential scope of the disease spread and the impact of different interventions. Here, I explain common quantitative principles underlying these models, including challenges in estimating the future rate of spread early in an outbreak. I also raise questions on how to retrospectively assess the value of different interventions.
Date, Time & Location: Thursday, March 31, 2016 at 11:30 am – 12:15 pm in Hendricks 3001
Speaker: Joshua S. Weitz, Professor, School of Biology Courtesy Professor, School of Physics Director, Interdisciplinary Graduate Program in Quantitative Biosciences Georgia Institute of Technology
Jan 27 (Wed) 11.30am – 12.15pm
CDC’s Ebola Emergency Response: contributions made and lessons learned
Brad Greening, PhD, National Center for Emerging and Zoonotic Infectious Diseases, CDC
Venue: Hendricks Hall 3001
In August 2014, CDC fully activated its Emergency Operations Center (EOC) to combat the threat of global spread of Ebola. The Modeling Task Force was embedded within the EOC scientific response structure to provide mathematical modeling expertise to Incident Management leadership. This task force provided insight into topics of concern such as the estimated number of Ebola cases that could be expected over time, as well as the estimated impact of interventions. Simple models were prioritized, which allowed decision makers to focus on what was needed to make specific choices in a data-limited environment. This talk will discuss key contributions of CDC’s Modeling Task Force during the height of the Ebola Response, as well as provide insight to the utility of mathematical modeling in a public health emergency setting. Finally, I will talk about some lessons learned during the response to assist mathematical modelers in future public health emergency settings.
Dr. Prateek Jain
In this talk – an Aerodynamic Designer and the Director of High Performance Computing at Gulfstream will explain how they design the world’s most coveted jet airplanes. They will also provide a behind-the-scenes look at the cutting-edge computing infrastructure that powers all these engineering efforts.
‘Implicit Functions for Image Based Modeling and Meshing’
March 27, 2015
Speaker: Dr. Reinhard Piltner, Georgia Southern University
Abstract: Extracting spatial information from sequences of medical images with the aim to generate finite element meshes for the simulation of the mechanical behavior of bones, blood vessels and implants can be very time consuming. Meshing for viewing 3D objects can be very different from meshing for numerical simulations of processes governed by partial differential equations. Utilizing the finite element method to get approximate solutions for differential equations, we need high quality meshes. Meshes with element distortions and the presence of very large finite elements next to very small ones usually will lead to unreliable results. In this presentation the option of using Radial Basis Functions (RBFs) for the meshing process and approximating signed distance functions will be discussed.
Design of the optimal control for chaos stabilization.
March 3, 2015
Speaker: Dmytro Dmytryshyn, Odessa National Polytechnic University, Ukraine
Abstract: We start with a brief introduction to the theory of quasi-dynamical chaos. Then an algorithm of chaos suppression will be introduced. Numerous examples of simulation and stabilization of chaos will be provided.
Correcting Errors in Linear Measurements and Compressed Sensing of Multiple Sources
October 12, 2012
Speaker: Alex Petukhov (UGA)
Fast el^1-greedy algorithm for sparse solutions of underdetermined linear systems
April 27, 2012
Speaker: A. Petukhov (UGA, joint with I. Kozlov)
The algorithms for finding sparse solutions of underdetermined systems of linear equations will be compared. Among those algorithms Orthogonal Greedy Algorithm, el^1minimization, Re-weighted el^1 minimization, el^1-Greedy Algorithm will be discussed. We also present a new fast algorithm combining ideas of fast implementation of OGA and el^1-Greedy Algorithm. The algorithm basic constructive block is one iteration of the standard interior-point linear programming algorithm. This algorithm combines computational complexity close to plain el^1-minimization with the efficiency of the sparse representations recovery approaching to the efficiency of the el^1-greedy algorithm.
Methods for diversity and overlap analysis in T-cell receptor populations
April 13, 2012
Speaker: Dr. Michael Seweryn, GHSU
Asymptotic analysis and numerical simulation of nonlinear Schrodinger equations with periodic potentials
March 8, 2012
Speaker: Christof Sparber (UIC Chicago)
We consider the time-evolution of quantum particles in periodic potentials under the influence of an additional slowly varying external field and possible nonlinear effects. This furnishes a natural two-scale problem, fundamental in several areas of quantum physics. We review some analytical results based on asymptotic analysis and also show how the insight obtained can be used in designing an efficient numerical method for the corresponding nonlinear Schroedinger equation.
Analysis of Parallel Replica Dynamics
March 1, 2012
Speaker: Dr. Gideon Simpson (University of Minnesota)
Parallel replica dynamics was first proposed by A.F. Voter as a numerical tool for accelerating molecular dynamics simulations characterized by a sequence of infrequent, but rapid, transitions from one state to another. An example would be the migration of a defect through a crystal. Parallel replica dynamics accelerates this by simulating many replicas simultaneously, concatenating the simulation time spent of the ensemble, as thought it were a single long trajectory. This leads to several numerical analysis questions: Is parallel replica dynamics doing what we think it is? For what systems will it be useful? How do we implement it efficiently? In this talk, I will thoroughly describe the algorithm and report on progress towards rigorous justification. Open questions will also be highlighted.