Dennis F. Kibler
Professor
Donald Bren School of Information and Computer Sciences
Donald Bren School of Information and Computer Sciences
PH.D., University of Rochester, Mathematics
OTH, University of California, Irvine
OTH, University of California, Irvine
University of California, Irvine
CS 414D
Mail Code: 3425
Irvine, CA 92697
CS 414D
Mail Code: 3425
Irvine, CA 92697
Research Interests
Computational Analysis of Genomes, gene regulation,motif discovery, gene expression clustering
Websites
Academic Distinctions
Appointments
Research Abstract
This research is being done in association with a number of faculty from Biological Science and Computer Science, most notably Pierre Baldi, Wes Hatfield, Rick Lathrop, Terry Long, Larry Marsh, Cal McLauglin, Ming Tan, and Suzanne Sandmeyer. There are a large number of interesting Machine Learning problems involved in understanding gene regulation and gene function, two of the most significant problems to be solved after the primary structure of the genome is known. The genomes that we focussing on are Chlamydia, Escherichia Coli, Saccharomyces cerevisiae, and Drosophila.
Currently our work centers on parsing and understanding the non-coding regions of the genome. We have developed a number of different techniques that depend on the type of information that is available. In particular for:
Gene Expression Data: We have developed an interactive clustering program that accepts gene expression data and determines corregulated genes. We have also developed several motif finding programs that accept the upstream regions of corregulated genes and detemine putative motifs. Depending on the users choices, the motifs can be represented as consensus sequences, probability matrices, or weight matrices. These choices affect the search complexity of the underlying algorithms. We are working on finding combinations of motifs as well as which representation is most appropriate.
Genome Sequence Data: We have a developed exploratory programs that rely solely on sequence data. The programs are capable of finding unusual patterns. These patterns are:
globally over-represented with respect to the genome
locally over-represented with respect to its neighbors
positionally conserved
micro-satellites
These patterns are interesting since they often correspond to biological meaningful subsequences, such as enhancers, repressors, activators, and promoters. For some patterns, no accepted biological function is known. We are developing additional program to find other types of patterns, such as LTRs (long term repeats) and CIRCE elements.
Promoter Region Data: We have developed a program to determine a weight matrix representation of a promoter region, given a set of common promoter regions. Our goal here is to evaluate various approaches to deriving weight matrices.
In general our goal is to develop programs that can use the available data to help determine biological significant substructures in genomes.
Currently our work centers on parsing and understanding the non-coding regions of the genome. We have developed a number of different techniques that depend on the type of information that is available. In particular for:
Gene Expression Data: We have developed an interactive clustering program that accepts gene expression data and determines corregulated genes. We have also developed several motif finding programs that accept the upstream regions of corregulated genes and detemine putative motifs. Depending on the users choices, the motifs can be represented as consensus sequences, probability matrices, or weight matrices. These choices affect the search complexity of the underlying algorithms. We are working on finding combinations of motifs as well as which representation is most appropriate.
Genome Sequence Data: We have a developed exploratory programs that rely solely on sequence data. The programs are capable of finding unusual patterns. These patterns are:
globally over-represented with respect to the genome
locally over-represented with respect to its neighbors
positionally conserved
micro-satellites
These patterns are interesting since they often correspond to biological meaningful subsequences, such as enhancers, repressors, activators, and promoters. For some patterns, no accepted biological function is known. We are developing additional program to find other types of patterns, such as LTRs (long term repeats) and CIRCE elements.
Promoter Region Data: We have developed a program to determine a weight matrix representation of a promoter region, given a set of common promoter regions. Our goal here is to evaluate various approaches to deriving weight matrices.
In general our goal is to develop programs that can use the available data to help determine biological significant substructures in genomes.
Publications
"Plateaus and Plateau Search in Boolean Satisfiability Problems: When to Give Up Searching and Start Again," with Steven Hampson. DIMACS Challenge, 1995.
Analysis of Yeast's ORF Upstream Regions by Parallel Processing, Microarrays, and Computational Methods. Steve Hampson, Pierre Baldi, Dennis Kibler, and Suzanne Sandmeyer. Tenth International Conference on Intelligent Systems for Molecular Biology ( ISMB-2000).
Combinatorial Motif Analysis and Hypothesis Generation on a Genomic Scale , with Yuh-Jyh Hu, Suzanne Sandmeyer, Calvin McLaughlin. BioInformatics (in press).
Detecting Motifs from Sequences with Yuh-Jyh Hu and Susan Sandmeyer, International Conference on Machine Learning 1999
Minimum Generalization via Reflection: A Fast Linear Threshold Learner , with Steven Hampson, Machine Learning Journal 1999.
Link to this profile
https://faculty.uci.edu/profile/?facultyId=2337
https://faculty.uci.edu/profile/?facultyId=2337
Last updated
12/17/2002
12/17/2002