UIUC Faculty Fellows Search

UIUC Faculty Fellows Abstracts for 2009-2010

Michael Dietze
Refined estimates of the eastern North American carbon budget: Multi-objective model calibration and data assimilation

There is a clear and growing demand for scientists to produce quantitative estimates of current and future impacts of climate change on terrestrial ecosystems that include robust estimates of uncertainty and risk. This project applies Bayesian statistical techniques to calibrate a sophisticated terrestrial ecosystem model, the Ecosystem Demography model (ED2.1), to thousands of sites simultaneously in eastern North America with multiple data constraints at each site derived from vegetation inventories, eddy-covariance, or remote sensing. Posterior parameter estimates are used to construct model predictive credible intervals and to conduct sensitivity/elasticity analyses. Next, data assimilation techniques are applied to estimate the current ecosystem state and fluxes for the study region. Finally, model forecasts will be generated for different climate change scenarios accounting for the uncertainties in the model, the initial conditions, and future climate. The successful completion of this project will result in a clear improvement of our understanding of the North American carbon cycle and the potential impacts of climate change in terms of an improvement in the spatial resolution, and improvement in the connection between models and data, and an improvement in the estimation of model uncertainty.

Chatham Ewing
Multi-Spectral Imaging and Analysis of Manuscript Materials

The aim of this project is to test the potential for using multi-spectral imaging, digital analysis, and post-processing to uncover lost texts within the manuscripts holdings of The Rare Book & Manuscript Library (RBML). We will focus on two fourteenth-century palimpsest manuscripts and a set of twentieth century letter-books that may benefit from this sort of analysis.

Yong-Su Jin
Optimal strain design for the production of ethanol from renewable biomass through computing elementary flux modes using a genome-scale stoichiometric model

Microbial cells have been utilized as a biocatalyst for conversion of biomass into value-added products. Metabolic phenotypes of naturally existing microorganisms are limited for various applications. Alternation of existing metabolic pathways or introduction of heterologous metabolic pathways for conferring beneficial phenotypes has been attempted. However, the metabolic phenotypes from the engineered microbial strains through metabolic engineering were often found to be suboptimal for commercialization, while successful in showing proof of the concept. The suboptimal phenotypes may be caused by uncharacterized interactions among metabolic pathways at the systems level upon perturbation of specific metabolic pathways. As such, we propose to employ a computational guide, enabling to predict global reconfigurations of genome-scale metabolic networks after specific genetic and environmental perturbations, for designing optimal strains for value added biotransformation. Specifically, we will focus on computing elementary flux modes (EFMs) from a medium-scale stoichiometric model in order to predict potential gene knockout targets improving yield of ethanol production from various sugars present in hydrolyzates of renewable biomass. A set of multiple knockout targets, eliminating the EFMs which are not favorable for high-yield ethanol production from the sugars, will be determined. Then, we will perform experimental validation for evaluating the effect of the identified knockouts on ethanol yields. Once the above mentioned strategy proves working, we will scale up the process using a genome-scale model through collaborations with NCSA scientists. The proposed study has both academic and industrial significances since the computation of EFMs from a genome scale metabolic model has not been reported and the outcomes of the proposed study, such as lists of knockout targets improving ethanol production, optimal ethanol producing strains, can be served as intellectual property.

Steven S. Lumetta
Enhancing GPU-based Supercomputing through Workload and Communication Optimization

GPU-enabled clusters, which offer huge reductions in cost and incredible power efficiency relative to microprocessor-based systems, may provide a viable means to exascale computing. Before the wider science and engineering community can effectively harness this potential, however, we must find methods to overcome the complexity barriers imposed by their inherently hierarchical communication systems and heterogeneous processing resources. I believe that by building on several recent advances, we can generalize some of the successes demonstrated by our dedicated application experts on these clusters so as to make the performance advantages available to a broader segment of the community.

Jian Ma
Algorithms and tools for mammalian genome reconstruction analysis

Data generated from numerous mammalian genome sequencing projects have provided a unique opportunity to use comparative genomics to computationally reconstruct the evolutionary history of our species and of other living species of placental mammals. The goal is to provide the trajectory of all the genetic changes leading to modern species, which may lead to novel biomedical discoveries. However, due to the diversity and complexity in modern mammalian genomes, computationally reconstructing the ancestral genome and the record of genetic changes leading to present day species is algorithmically extremely challenging. It is essential to combine all the complex evolutionary operations into a single, computationally tractable model. To capture the circumstances surrounding each of the major genetic shifts that occurred in the genomes, new and more efficient algorithms handling more complex scenarios need to be developed to solve the reconstruction problem at all different scales. Moreover, with the recent advancement in next-generation high-throughput sequencing technologies, an unprecedented amount of genomic data will become available soon. It is critical to make the software tools highly scalable to assist with making discoveries using huge amount of genomic data. The objectives of this project are to: (1) develop algorithms that handle complex evolutionary operations and develop software tools that can run on massive amount of genomic data; (2) reconstruct the ancestral mammalian genome and the evolutionary history on different branches; (3) use the reconstruction to study mammalian genome evolution. By taking advantage of the resources at NCSA that are not obtainable elsewhere, this project will create a set of novel software tools and make initial effort to build a foundational resource detailing the evolution of the mammalian genomes that can potentially be used by many other researchers to explore the genomes in an evolutionary context.

Junho Song
Rapid Decision Support for Hazard Responses by Cyberenvironment of Urban Infrastructure Networks

Society is increasingly demanding scientific accountability behind hazard mitigations and responses. Central to meeting these demands is to develop a cyberenvironment for infrastructure networks that can assess the effects of structural damage on the disruptions of residential/commercial activities, post-disaster responses and recovery efforts. Due to the complexity of the problem, risk assessment of lifeline networks is often performed by repeated computational simulations based on random samples of hazard intensities and component status. This sampling-based approach prevents rapid risk assessment and decision support for hazard responses based on online monitoring data. The main goal of the proposed project is to develop a non-sampling-based system reliability analysis method that enables rapid risk-informed decision support for hazard responses using a cyberenvironment of infrastructure networks integrated with network monitoring systems. The main theoretical framework of the research is the matrix-based system reliability (MSR) method, which allows rapid risk assessment and flexible probabilistic inference. The method has been further developed for multi-scale analysis of lifeline networks and for stochastic network flow analysis. In order to achieve the goal of rapid decision support based on online monitoring data, it is proposed to further develop the MSR method so that it can be integrated smoothly with network flow analysis algorithms. The method will be also generalized to obtain the joint probability distributions of monitoring data observed at multiple locations efficiently. This proposed approach is expected to account for the system effect of the network flow, the vulnerability of structures and hazard models through efficient system reliability analysis.

Jacob Sosnoff
Accelerometery in Wheelchair Propulsion

Nearly 70% of manual wheelchair users experience shoulder pain that negatively impacts their quality of life and independence. Although the pathogenesis of shoulder pain is multifaceted, propulsion mechanics (i.e., how one pushes the chair) appear to play a significant role. However, examinations of propulsion mechanics and shoulder pain have not yielded clear results. The majority of research on shoulder pain's relationship with wheelchair propulsion has been limited to laboratories due to technological constraints. It is thought that wheelchair propulsion in real life settings is distinct from that of lab settings and could exacerbate differences in wheelchair propulsion between those with and without pain. This proposal seeks to validate sensor technology (i.e., accelerometers) to allow for examinations of wheelchair propulsion in real life settings thus augmenting a larger project examining factors contributing to manual wheelchair users' shoulder pain. Overall, the project has the potential to develop techniques to identify wheelchair users at risk for developing shoulder pain and interventions to prevent shoulder pain, as well as influence manual wheelchair design. An integral part of this project is collaboration with the National Center for Supercomputing Applications, Peter Bajcsy and his Image Spatial Analysis Group, as he and the lab will provide expertise in sensor technology (i.e., 3D accelerometers). Results from this project will form grant applications to the National Institute of Health and the National Science Foundation.