2007-08 Research Projects

In 2007-08, five collaborative research proposals have been funded that focus on:

• Data mining methods for clinical and laboratory data
• Computational methods for rational drug design

The following projects are being supported by the University of Minnesota Rochester Biomedical Informatics and Computational Biology Program through its University of Minnesota/Mayo/IBM Collaboration Seed Grant Program.

1. Normalization in Global Mass Spectrometry Studies
2. Mining Genetic Determinants of Human Disease
3. Multi-Modality Data Mining: Accelerating Discovery
4. Petascale Computing Tools for Biocatalysis and Drug Design
5. Virtual Screening for Designing Selective ERK Inhibitors

Normalization in Global Mass Spectrometry Studies

Characterization of the human proteome is a resource of tremendous potential to biological research. Global proteomics via mass spectrometry is a powerful technology for study of the proteome; it has the potential to lead to a non-invasive screening mechanism of proteins in easily accessible body fluids. Using the iTRAQ isobaric tag labeling platform, four samples are subjected to mass analysis simultaneously. This is advantageous as it reduces experimental noise. However, there remains a need for normalization to remove systematic biases resulting from other steps in the process. In addition, functional analysis is needed to complete the biological story behind an experiment.

Investigators

• Yan Asmann, Mayo
• Robert H. Bergen, III, Mayo
• Jeanette E. Eckel-Passow, Mayo
• Ann L. Oberg, Mayo
• Terry Therneau, Mayo
• LeeAnn Higgins, U of M
• Vipin Kumar, U of M
• Gary Nelsestuen, U of M
• Michael Steinbach, U of M
• Baolin Wu, U of M

(back to top)

Mining Genetic Determinants of Human Disease

Integrating genomic information for computational identification of genetic determinants of disease promises to shed light on the causes and mechanisms of diseases in unanticipated ways. The objective of this project is to develop semi-supervised machine learning algorithms on graphs and supervised association analysis methods for the identification of clinically relevant SNPs that can serve as prognostic or diagnostic biomarkers. These algorithms will be applied to study genetic alterations in lung cancer, focusing on both neuroendocrine carcinomas of the lung and non-small cell lung cancer.

Investigators

• Jean-Pierre Kocher, Mayo Clinic
• Hugues Sicotte, Mayo Clinic
• Dennis Wigle, Mayo Clinic 
• Rui Kuang, U of M
• Michael Steinbach, U of M
• Vipin Kumar, U of M
• Richard Mushlin, IBM

(back to top)

Multi-Modality Data Mining: Accelerating Discovery

Advances in biomedical research areas such as genomics and imaging are providing new insights into functioning of the human body. These advances have also resulted in a rapid rise in the size and complexity of research data being collected by researchers. Traditional data analysis methods were not designed to deal with these very large datasets. The purpose of this project is to develop new data analysis tools that can deal with the analysis of large datasets quickly and efficiently.

The proposed data analysis tool will allow investigators to extract patterns out of a wide variety of large and complex datasets that otherwise may not be identified with currently available methods. These identified patterns can then guide the development of future studies. Importantly, these tools will allow investigators to maximize the usefulness of information obtain from their data.

Investigators

• Kelvin Lim, U of M
• Vipin Kumar,U of M
• Monica Luciana, U of M
• Michael Steinbach, U of M
• Tim Mullins, IBM
• Richard Mushlin, IBM

(back to top)

Petascale Computing Tools for Biocatalysis and Drug Design

An inter-disciplinary research strategy to create next-generation computational tools for biocatalysis and drug discovery. The plan is to develop radically novel application software and cyber-infrastructure that enables collaborative research in a virtual “problem-solving environment." At the core of this problem-solving environment will be innovative multi-scale modeling methods poised to take advantage of emergent petascale computing resources, the results of which feed into a network of linked databases for drug discovery.

There are three specific aims:

1. To develop a new fully quantum mechanical molecular force field for virtual screening and biocatalysis simulations,
2. To create a prototype networked quantum database with data-mining tools for drug discovery and lead optimization, and
3. To apply the quantum force field and database to driving problems of catalytic mechanism and rational drug design. 

Investigators

• Darrin York, U of M
• Yuan-Ping Pang, Mayo Clinic
• Carlos Sosa, IBM

(back to top)

Virtual Screening for Designing Selective ERK Inhibitors

In the U.S., mortality due to cancer has already surpassed mortality due to heart disease in citizens under 80 years of age. Limited success of current treatments for cancer underlines the importance of more efficient, less toxic, treatments.  Understanding cancer at the molecular level will lead to better tolerated therapies with fewer side effects. Most human cancers exhibit deregulation of numerous cellular signal pathways that are involved in cellular growth or death. The mitogen-activated protein (MAP) kinases are major signal transduction routes to transfer messages from the cell surface to the nucleus and have been implicated in uncontrolled cell proliferation and tumor growth and therefore are an important therapeutic target.

The Hormel Institute (U of M) and IBM (Rochester & Minneapolis) propose the following specific aims:

• Aim 1
  Using IBM Blue Gene, plus multiple computational software packages, which will be installed at the Hormel Institute computational facility, we will identify small molecules that show high affinity for selected binding of specific mitogen-activated protein (MAP) kinases.
• Aim 2
  We will test and verify the selective inhibition of the small molecule to these MAP kinases.  Completion of these aims will yield selective, small-molecule inhibitors of selected MAP kinases that can be used to study signal transduction and will test the hypothesis that selected MAP kinase proteins can be used as anticancer drug targets. These inhibitors will have extensive commercial use (e.g., PD98059 is already available to inhibit upstream MEK1).

Investigators

• Ann Bode, Hormel Institute
• Zigang Dong, Hormel Institute
• Angelo Pugliese, Hormel Institute
• Carlos Sosa, IBM

(back to top)