Wayne State University

AIM HIGHER

Wayne State University

High Performance Computing Services

Research

This space will be used to showcase both the research being done using the Grid computing system and the scholars that are conducting it.

If you are currently using the Grid and would like to have information about your work published here then please write to Aragorn Steiger at aragorn@wayne.edu.
To view the list of published research done with the use of the Grid, please follow the link to the Publications page here.
Anupam Bhattacharjee

Testing network analysis algorithms in parallel and non-parallel modes.

The Chernyak Group

Our research mainly concerns about electronic structure of conjugated systems, especially excited states properties. We mainly use (compile and execute) Gaussian 03 package and some home made codes (Fortran 77) on the Grid to compute the above mentioned structure and properties. These calculations are computationally intensive and need stable workstations(nodes) and smooth network for data transport between nodes and our control desktops.

The Dombkowski Group
NeMo: A network modeling server for biological pathway analysis.

High-throughput genomic and proteomic technologies have recently enabled unprecedented views of gene and protein expression in cellular systems. These methods provide a staggering amount of data, even in modest experiments, and effective interpretation of the data is highly dependent on advanced computational techniques to analyze, manage, and visualize the results. The integration of computational biology and “omics” technologies has spawned the systems biology approach towards characterizing cellular events on a broad scale, including elucidation of complex networks of genes and proteins, and their regulatory mechanisms. These approaches require numerous, highly specialized computational tools and the associated skills, which are generally out of reach to most investigators. An important objective towards providing systems biology capabilities to the general research community is the development and integration of cohesive, web-accessible tools and databases that can be utilized by non-specialists. The proposed research will develop a network modeling system comprised of a high-throughput genomic and proteomic database and associated computational tools to enable predictions of cellular networks. Initially, the database will be populated with data and models for one of the most important signaling pathways in biomedical research: the PI3K/Akt/mTOR pathway. The computational framework developed in this research will be extensible to facilitate inclusion of other signaling pathways and data types.

Our long-term goal is to identify the topography of regulatory networks critical to cellular processes and disease progression through the development of computational methods and their application to high throughput genomic and proteomic data. The objective of this application is to develop a database and tools that will provide models of transcriptional regulation controlled through the PI3K/Akt/mTOR cascade using a systems biology approach. Subsequent work will focus on extending this approach to other signaling cascades along with the integration of other high-throughput data types such as ChIP on chip and microRNA array data.

NeMo.doc
Radu Florea

Software for computational fluid dynamics, specifically involving chemical kinetics.

The Friedrich Group

Phylogenetic reconstruction is the science of taking traits we observe in modern (and sometimes fossil) organisms and using the distribution of these traits to determine the evolutionary relatedness of the organisms. As part of the AToL: Diptera (FLYTREE) project, we are investigating the approximately 250 million year history of the true flies using molecular sequence data gathered from 43 species representing the full breadth of Dipteran diversity. In addition to probing the phylogeny of the flies, we are using this data set to answer broader questions regarding the phylogenetic utility of several classes of data. Utilizing parallel computing approaches, we are comparing the performance of mitochondrial genome encoded genes versus those found on the nuclear genome and evaluating various codon position based partitioning schemes. Furthermore, computationally derived divergence time estimates are allowing us to take a nuanced view of how phylogenetic utility can change over time.

Dr. Domenico L. Gatti

Molecular dynamics simulation of a very large system comprised of ~600,000 atom representing a membrane protein (cytochrome oxidase), which is one of the key human enzymes necessary for respiration. Both the wild-type and several inactive mutant proteins will be simulated in order to understand the pathogenesis of some neurological and muscular disorders.

The Khalil Group
Vehicle Crashworthiness and Occupant Protection

Students in ME 8020 were assigned a term paper on developing Finite Element crash models of cars and trucks. These models have about 500,000 degrees of freedom (500,000 equations) and require a solution for 200 time steps. We use LS DYNA as a solver which is part of “PACE”.

The Lipovich Group

Drug addiction is a fundamental threat to public safety and health. Its pathobiology has therefore been under intense study. The nucleus accumbens (NAc) is a brain region that plays a key role in normal reward learning and in addiction. Gene expression microarrays have been employed to elucidate the molecular basis of NAc neuroplasticity in addiction, revealing a complex landscape of distinct and drug-specific gene expression patterns in animal models and human postmortem brain. While such microarrays predominantly measure mRNA levels of proteincoding genes, major recent studies demonstrate that non-protein-coding RNA classes, including microRNA and long noncoding RNA (lncRNA), are abundantly expressed in the brain within specific spatiotemporally restricted contexts, and can directly regulate protein-coding genes. Nevertheless, nothing is known about human brain lncRNA expression in drug addiction. The proposed project will fill this important gap in our knowledge of brain gene expression and function. Insights emergent from the proposed analysis of lncRNA expression in the addicted and normal brain will further our understanding of addiction mechanisms and may facilitate the development of anti-addiction therapies.

Specific Aim 1 will construct a comprehensive catalog of human lncRNA genes, and will build upon that catalog to design a custom genome-wide lncRNA expression analysis microarray. This is a fundamental necessity, since there is no universally accepted reference annotation of the human lncRNA transcriptome, and since prefabricated microarrays exist to profile proteincoding and microRNA transcripts but not lncRNA. Specific Aim 2 will use the arrays created in Aim 1 to identify lncRNAs differentially expressed in the NAc of cocaine abusers, heroin abusers, and pair-matched controls, and will validate the microarray results using RT-PCR. The outcome, a catalog of lncRNAs whose NAc expression is altered significantly by chronic drug exposure, will provide a foundation for functional studies of lncRNA action mechanisms in drug addiction. Specific Aim 3 will begin to elucidate the regulation of the differentially expressed lncRNAs by combining public transcription factor binding site datasets with lncRNA genomic mappings, thereby identifying neurally-expressed transcription factors with potential binding sites at promoters of specific lncRNA genes. Fitting specific lncRNAs into known transcription regulatory networks will begin to elucidate the regulation of lncRNA expression, potentially highlighting signaling pathways amenable as therapy targets for drug abuse.


Lipovich_NIH_March08 project summary.pdf
The Loeb Group
Systems Biology - “The ability to obtain, integrate and analyze complex data from multiple experimental sources using interdisciplinary tools”.

Epilepsy is a disease of recurrent seizures that affects up to 1% of the world’s population. Yet it remains one of the least understood human disorders in the most complicated of human organs, the brain. As a means to understand and develop improved diagnostic and treatment strategies, we have formed an interdisciplinary collaborative project that uses the power of systems and computational biology to understand human epilepsy through its electrical, anatomical, and molecular features.

This work has lead to the discovery of a ‘final common pathway’ of genes that are consistently induced at human epileptic foci and now serve as invaluable molecular markers and drug discovery targets. We have recently been funded by the President of Wayne State University to build a user-friendly database of human epilepsy that links electrical, anatomical, and molecular features of human epilepsy that brings the latest advances in bioinformatics workflow. All of this work truly reflects the human condition as it is made possible through our far-sighted patients that have allowed us to study portions of their brains removed for the treatment of medically refractory epilepsy.

The Potoff Group

Our research involves force field development of compounds that are building blocks to variety of complex molecules ranging from chemical warfare agents to biological molecules like protein, lipids and carbohydrates. We use Monte carlo methods to evaluate the performance of the models and also reproduction of macroscopic quantities which are experimentally measurable. There is also use of NAMD molecular dynamic package to evaluate the non equilibrium properties of materials of interest.

The Rajlich Group

The goal of the project we conducted using the Grid is to reveal hidden dependencies in open source systems using invariants generated by Daikon invariant detector. Using the Grid we performed a case study on Apache FtpServer open source project and revealed 3720 potential hidden dependencies. All the invariants used in this case study were created using the Grid. The results will be included in our next paper. We would like to extend the study and we are currently searching for another project to be analyzed.

The da Rocha Group
Drug Delivery using Water-in-HFA Microemulsions: Surfactant Design Through Computer Simulations and Atomic Force Microscopy

This research is aimed at developing a fundamental knowledge of the interfacial properties of the bare and surfactant-modified hydrofluoroalkane|water interface (HFA|W). The goal of our project is to design efficient amphiphiles capable of forming and stabilizing aggregates of water dispersed in HFA, for use in pressurized metered-dose inhalers. Such novel formulations are expected to deliver biomolecules via pulmonary route. Therapeutic agents, including antibiotics and anticancer agents, can be incorporated in the aqueous phase and create new perspectives for the treatment of medically relevant diseases.

Within this context, we used GRID to run atomistic molecular dynamics computer simulations in order to determine the thermodynamic and microscopic interfacial properties of the HFA|W interface. In addition, ab initio calculations using Gassuan 03 were performed in order to study the solvation of the surfactant tail in HFA propellants.

daRochaGroup.doc
The Sarhan Group

Our work on the cluster is basically simulations for "scalable media streaming systems".

The Schlegel Research Group

Our lab is involved in both the development and the application of new methods in ab initio molecular orbital methods. Many of our applications of quantum chemical calculations are in direct collaboration with experimental groups in order to maximize the benefits of our studies. Our research is diverse, touching on topics in strong field chemistry, molecular dynamics, bio-organic interactions, materials science, as well as inorganic transition metal complexes and redox chemistry. A complete list of our group’s publications is available at: http://chem.wayne.edu/schlegel/

The Schlegel Research Group.doc

Schlegel Group research image 1 Schlegel Group research image 2 Schlegel Group research image 3
Alan Sebastian

Using the Gaussian program to do calculations complimenting research concerning the fragmentation of biological molecules, starting with simple amino acids. Also, single threaded code to calculate the transport of electron swarms through various gas mixtures, currently being rewritten to make use of shared memory multicore/multithreaded machines.

The Tainsky Group
Proteomic Approaches to Cancer Diagnostics Using Antigen Microarrays

Discovery of new biomarkers for early cancer diagnostic, using the R language on the WSU Grid facility for sophisticated statistical analysis of high-throughput protein microarray data.

The Verani Group

The Verani Group develops coordination chemistry aiming at amphiphilic materials and metallodrugs. Computational calculations are used to model electronic properties in such systems.

http://chem.wayne.edu/veranigroup/
The Wada Group

Estimating economic fluctuations; and testing and simulating economics models.

The Wadehra Group

Using the computational facilities of the Grid, Dr. Alan Sebastian and I have done a number of new and interesting calculations in atomic physics.


The Cheng-Zhong Xu Group

We are working in a project on developing a failure-aware resource management system for high-performance computing. We have been working with Philip Sokolowski and Michael Thompson, and conducting experiments on WSU Grid since 2006. As the first step of our project, we have analyzed the characteristics of failure occurrences using the failure and performance traces collected from the Grid. Based on this information, we have proposed a proactive failure management mechanism to forecast failure occurrences in the future. The research result and experimental data have been presented in two papers accepted by ACM/IEEE SC'07 and IEEE SRDS'07.

Now, we are working on the 2nd step of our project. By using results of failure prediction, we are developing a resource management mechanism that schedule jobs based on the reliability status and performance of grid nodes. Currently, we are conducting some experiments on the Grid. We will integrate the results into a paper to be submitted to this year's SuperComputing conference SC'08. We are working with C&IT to find compute nodes to evaluate the performance of our mechanisms and applying our approaches to facilitate system management in the Grid.

failure management.doc
Computer Center Services * 5925 Woodward Avenue Detroit, Michigan 48202