Brendan Mumey - US grants

DBI 9978582; SGER Computational Neuroscience Problems with application to Microarray Expression Analysis

The objective of the proposed research is to develop and implement computational
approaches for determining the mechanisms through which information is represented
and processed within neural systems. The project will then examine the hypothesis that this problem is similar to studying gene expression patterns using DNA microarrays in order to learn about the underlying regulatory networks that presumably govern gene expression. Thus, the proposed overall intent is to adapt methods developed in the neural domain to analyze gene expression data from microarray experiments.
The premise is that the scientific problem of neural mapping and the problem of using DNA microarrays to understand gene regulatory networks share some interesting similarities. The underlying assumption is that both systems, gene product regulation and neural activity are governed by a control network-understanding in greater depth and subsequent computational modeling of the neural system could, when applied to the network control analogy in gene expression provide valuable insights and ultimately the basis of a new set of informatics tools designed to better analyze microarray gene expression data.

EIA-0129895 -John P. Miller-Montana State University-Algorithms for real-time decoding and modulation of neural spike trains-A grand challenge in neuroscience is to understand the biological basis of information processing in nervous systems. Three major goals facing sensory neuroscientists are a) to understand how sensory information is encoded in the activity patterns of neural ensembles, b) to understand how those activity patterns are decoded by cells at the subsequent processing stages, and c) to understand how computations (e.g. visual pattern recognition or oriented motor responses) are carried out on that decoded information. Two major goals of the research proposed here are a) to develop a formal, general approach toward achieving those goals, and b) to test and refine that approach by characterizing the functional organization and neural encoding scheme of a simple sensory system. These goals will be achieved through the development of a data-driven model of the system. The model will be formulated in terms of information processing units and information channels, rather than in terms of individual neurons. That is, the functional units in the model will be operators that carry out specific, independent computations (or information transformation operations) at a specific processing stage in the test nervous system, and the channels through which information is passed between these functional units will correspond to information channels in the Shannon sense.

Physically Aware Agile Networking
PI: Brendan Mumey

This research project considers agile constraint-based routing and wavelength assignment (RWA) strategies to mitigate physical layer impairments that affect the end-to-end performance of next generation all-optical networks. The RWA methods being developed take into account changing traffic patterns, resource availability, and the behavior of the underlying physical layer under transient conditions to minimize impairments and assure that Quality of Service (QoS) requirements are achieved. Physical layer network impairments that affect end-to-end path performance include switching transients, amplifier gain fluctuations, and channel cross talk. Temporal and geographic variations in network traffic can also affect network performance. A general framework for studying the cumulative effects of such impairments and traffic variations is being designed and conducted using the Montana State University optical network testbed. This work requires a combination of modeling, simulation and validation steps. Based on this new framework, state-of-the-art physically aware RWA algorithms are being designed, implemented and tested.

Broader Impact:

The outcomes of this research will become an important component for future cyber infrastructure by exploring methods to manage the underlying network that are consistent with the new application-driven network demands. This project will integrate teaching and research activities in electrical and computer engineering with computer science, bringing together faculty and students from both disciplines to address system-level design issues spanning both disciplines. The project team will engage undergraduate students in the research through interdisciplinary senior design courses and individual research projects. Outreach to minorities will be facilitated through the well-established programs at MSU that engage Native Americans in the sciences and engineering.

This funding establishes a new CISE Research Experiences for Undergraduates (REU) Site at Montana State University focused on networking and networks with applications for sustainability. Each summer a cohort of undergraduate students will participate in a ten-week summer research program at the host institution. The project includes mentoring by the experienced computer science faculty members, technical seminars and workshops, student presentations, and field trips and other professional development opportunities. The REU Site program will target high-quality undergraduate American Indian students, students from tribal colleges, and students from small colleges with limited resources. Recruitment efforts will focus on computer science and engineering students as well as students from environmental science and other related disciplines. This site is co-funded by the Experimental Program to Stimulate Competitive Research (EPSCoR).

The intellectual merit of this project lies in strong research basis and the expertise of the faculty. The projects are in research areas that are current and address national priorities such as networking and sensor networks. The students participate in a full range of research activities from preparing research literature reviews to production and dissemination of research results. The project has the potential to add to the research base of networking systems.

The broader impacts of the project include providing a quality research experience to undergraduate students, particularly students from underrepresented groups. The project team is committed to including under-represented minority students in their research. Thus this project has the potential to produce new computer science graduate students and faculty members and to advance discovery and understanding while promoting learning.

As DNA sequencing continues to improve in quality and decrease in cost, the volume of genomic data continues to increase at a geometric rate. In particular, multiple genomes from the same or related species are actively being acquired; this has lead to the new research field of pangenomics. Pangenomics sheds new light on how organisms adapt to their environment and what genomic features vary or stay the same within related species. This project deals with some of the computational challenges associated with efficiently storing and querying large pan-genomic data sets. New and enhanced data structures will be developed that efficiently describe the genomes of closely related lines or species and permit fast information retrieval for common activities such as sequence searching. The investigators will develop software tools that will be made freely available to the general scientific community and test and validate these tools using several important species, including Saccharomyces cerevisiae (a yeast species), Arapidopsis thaliana (thale cress, a small flowering plant) and Medicago truncatula (barrel clover, a small legume).

This project will initiate work on a next generation of bioinformatics algorithms that can exploit the increased information content available from multiple accessions and intelligently use the data for unbiased, species-wide analyses. The current trajectory of next generation sequencing improvements, including falling costs and increased read lengths and throughput, ensure that multiple genomes per species will be routine within the next decade. The researchers will investigate improvements to graphical data structures such as deBruijn and string graphs, used to represent pan genomes and develop associated algorithms for querying pan-genomic data, building on data structures such as the FM-index. Saccharomyces cerevisiae, Arapidopsis thaliana and Medicago truncatula will be the principal model organisms for this study, listed in order of complexity. Multiple genomes of each of these organisms have been sequenced so they are good candidates for pangenomic study. In particular, this project will complement existing work on symbiotic relationships of M. truncatula. This work will also shed light on the impact of reference bias in genomic analysis and provide alternative routes to avoid it. It will provide a practical method for pan-genomic sampling that provides a single FASTA sequence that minimizes reference bias in downstream analysis. This research will lead to new software tools that will be made available to the community and will be used to engage students in bioinformatics research and educational activities.

The "Biofilm Resource and Information Database (BRaID)" project is a collaboration between the Center for Biofilm Engineering (CBE) at Montana State University, Bozeman, MT and the Gianforte School of Computing at MSU and the National Center for Genome Resources (NCGR), Santa Fe, NM. Biofilms are microbial communities that have important impacts on water quality, food production, energy-making processes, and various aspects of human and animal health. BRaID serves as a community resource, containing both data and the metadata (where or how data was collected, processed, and interpreted). Some examples of metadata include: a geographical location, such as the Bozeman Water Reclamation Facility; microenvironments such as water chemistry; environmental factors such as changes in temperature, perturbations such as adding corrosion inhibitors; and interventions such as antibiotic treatments. BRaID has utility to a broad range of stakeholders in basic and applied research, as such it is expected to enhance the competitiveness of US industry, basic research, and defense, as well as assist with advancements in human health, environmental health, and energy production. Outreach activities will include (1) dissemination of research outcomes to local community members and industry stakeholders at conferences and legislator engagements; (2) involvement of underrepresented Native American students from high schools and tribal colleges in summer research activities as well as undergraduate researchers at MSU-Bozeman; and (3) educational activities in undergraduate and graduate courses at MSU, as well as summer internship, outreach and training activities at the NCGR.

Awareness of the importance of microbial biofilms - microorganisms growing as aggregates attached to surfaces, interfaces, or each other - has exploded in the past decade. Diverse microbial communities grow as biofilms in such settings as dental plaque, heart valves, natural watercourses, wastewater treatment processes, cooling systems, oil and gas pipelines, and persistent infections. Biofilms can have significant impacts on human health, industrial productivity, and natural resources, spanning natural, engineered, and medical systems. Despite the importance of biofilms, there is no public web portal and database dedicated to storage, analysis, and communication of biofilm-specific data. Part of the reason for this is the sheer breadth of data types used to describe biofilms, including nucleotide sequences, images, video, chemistry (water, surface, and extracellular polymer matrix), remediation history, clinical outcomes, and geospatial data. BRaID will complement existing microbial databases by developing a data representation paradigm that solves the problems posed by the challenge arising from the unique combination of data, and will implement new metrics and algorithms for the formal description and analysis of biofilms data. This project aims to provide the premier resource that offers users the ability to ask complex questions (including those currently impossible to answer), permitting quantitative comparisons among biofilms, supporting remediation efforts, and enabling predictions about outcomes. Results of the project can be found at: http://ncgr.org/braid.

The objectives of Montana State University's (MSU) Computer Science (CS) Department REU project are 1) to expose undergraduate students to real-world, innovative and interdisciplinary algorithms research founded in modern software engineering techniques; 2) to encourage more undergraduates to continue their academic careers and seek graduate degrees in computer science and related disciplines; 3) to develop research skills and improve communication and collaborative skills; and 4) to foster collaboration and entrepreneurship in the development of prototype software through our innovative Software Factory environment. The REU program will aim to broaden impact and participation by recruiting high-quality undergraduate students from regional tribal colleges, students from small colleges with limited resources, as well as women engineering students to increase the gender and cultural diversity of students. The REU projects encourage students to take a systems-level perspective in solving problems from an ever-growing number of diverse commercial, educational and environmental domains. This REU will increase student awareness of cultural diversity and bolster academic confidence and interest among underrepresented groups to aid in retention and advancement, as well as preparing students in real world development environments. The program will leverage highly successful MSU campus-wide and College of Engineering programs already in place to recruit and retain underrepresented students from Montana tribal colleges.

By using a realistic software development environment - the Software Factory, the PIs will heighten undergraduate student participation in innovative research of algorithm development. Project topics have been carefully selected by each mentor to make sure they include a significant development component. It seeks to increase the number and diversity of students entering graduate programs and research careers in computer science and related fields. The program will target a variety of topics in interdisciplinary research aligned with every mentor's area of expertise. The students will acquire key general research skills, learn how to design algorithms and develop and use tools to tackle complex problems.