Research

 

DARPA LogoNSF LogoNIH Logo

The Biomathematics Research Group (BRG) at UGA explores mathematical and computational models that connect genes to ecology. The BRG is producing advances in the following areas:

  • Multi-scale analysis of infectious disease: We study mechanisms that connect multiple scales, from milliseconds through evolutionary time, and from quantum interactions through continental dynamics of infection. This endeavor requires advances in multiple areas, as described below. We have been able to produce advances in every single aspect, and we are capable of integration of all these dimensions. 
  • Data harmonization: We are able to represent heterogeneous and complex data sets of arbitrary size with a reduced set of data primitives (well-defined mathematical objects that are the building blocks to represent reality). Our analysis pipelines only consume data primitives, and only produce data primitives. The ultimate goal is to design and implement an agent capable of automated knowledge discovery.
  • Machine learning: We use machine learning on a regular basis, but we are also developing new algorithms that improve the performance of artificial neural networks as compared to the state of the art.
  • Adaptive learning: The training of interdisciplinary scientists poses tremendous challenges. This is particularly true when teams are comprised of people with heterogeneous backgrounds. We have developed technology that minimizes the cognitive overhead to train individuals, and to integrate teams into a research project.
  • Epidemiology of asymptomatic carriers: We study the effect of asymptomatic hosts in the dynamics of malaria transmission. The results can be extrapolated to other diseases. 
  • Dispersal of vectors: We study methods to model the dispersal of mosquitoes in a heterogeneous landscape dominated by species distribution, climate, and vegetation cover.
  • Physiology through Telemetry: We are able to identify when a host is infected before the onset of symptoms occurs. We accomplish this via high-frequency measures of accelerometers, blood pressure, ECG, and temperature. 
  • Cellular models of immune interaction: We are using flow-cytometry data and cytokine information during a malaria infection to model cellular-level interactions between the adaptive and innate immune system, and healthy and infected red blood cells.
  • Multi-omic integration: We are able to analyze transcriptomic data, and integrate it with proteomics, metabolomics, immunomics, and other -omic technologies. Using this type of integration we have been able to determine factors that confer resilience against an infection.
  • Computational Drug Design: Given a gene regulatory network (that we can reconstruct from a time series of transcriptomic data), we can identify the most sensitive elements of the network, and target them with molecular docking studies against of databases of drug-like molecules.

Our currently funded projects are:

  • (PI Gutierrez) ALICE (Adaptive Learning for Interdisciplinary Learning Environments, 2016-2018, $299K), NSF award #1645325 : ALICE is a Web-based information system that generates individualized development plans, according to previous experiences and current challenges. Furthermore, ALICE is designed to connect lexias from multiple subject matters, thus bypassing disciplinary barriers that in many cases are artificial. The principles behind ALICE are generalizable, and hence it has the potential to be used in K-16, graduate, and continuing education.
  • (Co-PI Gutierrez, PI Galinski) Technologies for Host Resilience (2016-2019, $1.9M UGA out of $6.5M) - Host Acute Models of Malaria to study Experimental Resilience (THoR's HAMMER), DARPA contract contract #W911NF-16-C-0008, 2016-2019. This project will explore the molecular mechanisms of resilience, susceptibility and resistance of non-human primate hosts when challenged with a malaria infection.

Our past funded projects are:

  • (Co-I Gutierrez, PI Galinski) Malaria Host-Pathogen Interaction Center (2012-2017, $19M), MaHPIC, NIAID contract #HHSN272201200031C, 2012-2017. PI Mary Galinski. MaHPIC involves the multidisciplinary study of malaria infections, immunity and pathogenesis of P. falciparum, P. vivax and P. knowlesi in the context of host-pathogen interactions, in humans and nonhuman primates, using a systems biology approach. Three nonhuman primate malaria species will be studied: P. coatneyi to model P. falciparum, P. cynomolgi to model P. vivax, and P. knowlesi, a monkey malaria species that has been causing illness and cases of death in humans in Southeast Asia.
  • (Co-I Gutierrez, PI Herrera) International Centers of Excellence in Malaria Research (2011-2017, $5.5M) - Center for non-Amazonian regions of Latin America - CLAIM, NIAID cooperative agreement #U19AI089702-01, 2010-2017. PI Sócrates Herrera. CLAIM is divided into three projects: Project 1 is evaluating the diversity of the ecology and parasite populations related to the epidemiology and clinical findings to establish a scientific framework that supports the development of new intervention strategies for malaria elimination in non-Amazonian areas of Latin America. Project 2 is addressing major gaps in understanding of the ecology, behavior, vector potential, and control of Anopheles malaria vectors to guide the development and implementation of more effective integrated vector management (IVM) strategies of National Malaria Control Programs (NMCPs). Project 3 aims to determine the clinical outcomes and their association with parasite and host features of malaria-infected individuals living in non-Amazon regions of LA with different intensities of malaria transmission. 

We are currently interested in the following projects:

  • Bioclassification: This research project involves the characterization of biological objects (RNA, DNA, proteins, brains, time series, etc.) through a minimal set of data primitives. This system produces different types of data structures and feature vectors, e.g. higher order moments, knot invariants based on Gauss integrals, etc. and then applies a battery of pattern classification algorithms (e.g. Multiple Discriminant Analysis (MDA), Principal Component Analysis (PCA), clustering techniques, neural networks, etc.) for the purpose of automated object classification. We built a web front-end to a high-performance relational database for large data sets available at: www.bioclassification.org

Examples of past research projects are:

  • Trojan chromosomes: The Trojan chromosome hypothesis states that it is possible to achieve local extinction of exotic species through the constant environmental introduction of sex-reversed organisms, which are produced via phenotypic and genotypic manipulations.
  • Sonification: This project explores the possibility of hybridizing sound and visualization to represent complex data sets to harness the human ability to find patterns. For instance, a common problem in data visualization is the representation of higher dimensions. A surface plot could represent three dimensions. You can always add two additional dimensions via color and sound. First the deceivingly simple "color" option. If you use MATLAB, you know that you have several options for color maps that simply represent the Z axis... but what if your color is not related to the Z axis, but instead to other dimension. The second option is sonification of the image, that is to say, each point in your 3-D plot has a related sound that presents the 5th dimension of your data.
  • Invasive Species Diffusion Model Prototype: Spatial spread of invasive species can be modeled with a system of advection-diffusion-reaction partial differential equations. Such a model was used to predict the advance of an invasive front of Channeled Apple Snails (CAS) and estimate population densities over time across the entire spatial domain. http://www.youtube.com/watch?v=joR4Rtzm12c&feature=PlayList&p=3A96AC32A930CE17&index=0&playnext=1
    This list includes: (i) video demo of an invasion process of CAS in Lake Munson, Tallahassee, FL, (ii) video-manual to extract information from a map, (iii) video-manual to enter information into Scilab, (iv) video-manual to produce a invasion model video for a user-defined spatial domain.
     
  • Analysis of Financial Time Series: In this project we studied a Terabyte-size database of quotes and trades for 38 symbol in a period of 4 months, and developed algorithms for data filtering that eliminate the delay associated with moving averages. The techniques used have to do with Fourier analysis, complex extensions of real functions, and use of fat-tail distributions to estimate variations. This project resulted in a proprietary trading algorithm used by Level 3 Data.
     
  • Literatronica: An artificial intelligence engine that makes an approximation to a Hamiltonian cycle in a weighted graph with weights that vary in time according to a coupled set of ODEs, to achieve adaptivity in my own literary works. Adaptive digital narrative works reconfigure themselves for the reader, leading to a potentially unique read every time. Adaptive books cannot be reproduced on paper except, perhaps, as a reading path at a given moment. The solution of a multi-terminal network flow on such a graph solves the optimization problem of minimizing hypertextual attraction (a measure of narrative continuity) and hypertextual friction (a measure of the risk of loosing readers' attention). Available at: www.literatronica.net