NMR Resonance Assignment and Microarray Data Analysis (May 2001)

The Computational Protein Structure Group in the Computational Biology Section of ORNL's Life Sciences Division has made significant progress in the research areas of NMR resonance assignment and microarray data analysis.

We developed a computer program and submitted a paper for the problem of NMR resonance assignment, which is one of the key steps in solving an NMR protein structure. The assignment process links resonance peaks to individual residues of the target protein sequence, providing the prerequisite for establishing intra- and inter-residue spatial relationships between atoms. The assignment process is tedious and time-consuming, which could take many weeks. Though there exist a number of computer programs to assist the assignment process, many NMR labs are still doing the assignments manually or semi-manually for quality reasons. We designed a new computational framework for automating the assignment process, particularly for backbone resonance peak assignment. We formulate the assignment problem as a constrained weighted bipartite matching problem. The formulation provides a natural framework for incorporating all available information into the assignment process. We have implemented the algorithm and tested it on four proteins with both real and simulated NMR peaks. The promising results indicate our method has made further progress in fully automated peak assignment. This work was submitted for publication to a special issue of "Intelligent Systems in Biology" in IEEE Intelligent Systems & Their Applications at the end of May 2001.

We developed a computer program to analyze microarray data and applied the program to study the gene expression to chitin in Arabidopsis. One of the most powerful tools to investigate gene function and pathway is DNA microarray. Massive microarray data on gene expression have been generated. Although some computational tools have been developed to analyze the data, they generally do not meet the needs of experimentalists well due to their limitations in algorithms, utilities, and software interfaces. To address these limitations, we have developed a computer program for clustering gene expression profiles of microarray data using a new approach based on the minimum spanning tree algorithm. Our preliminary studies show that the program produces good results in benchmark tests. It also runs fast, and only takes seconds for analyzing a set of microarray data. The program offers varieties of options in addition to an easy-to-use default setting. It is integrated with a user-friendly JAVA interface. We have applied this program to a project in collaboration with Gary Stacey's group at the University of Tennessee and Shauna Somerville' Group at Carnegie Institution on the analysis of gene expression and upstream regulatory regions of plants. We clustered the chitin-responsive ESTs according to their expression profiles. We also searched the upstream regions of the corresponding genes, and carried out a comparative promoter analysis of the genes in the same cluster. We identified some interesting conserved motifs, including several known binding motifs such as W-boxes.

The study of NMR resonance assignment is funded by DOE OBER (a project titled "Structures of DNA-Repair Proteins: A New approach Combining Computational Modeling and NMR Spectroscopy"). Microarray data analysis is funded by LDRD at ORNL (a project titled "Computational Inference of Regulatory and Metabolic Networks").

Contact: Ying Xu, 865-574-7263 or xuy1@ornl.gov

Funding Source: DOE-OBER (KP)