GUEST SPEAKER:          Simon Kasif
                        University of Illinois at Chicago &
                     Johns Hopkins University

TALK TITLE:      DATASCOPE: A Computational Learning Approach for Whole-Genome
Interpretation

Abstract: In 1995 the first complete genome, H. influenzae, was sequenced
and published by The Insitute for Genomic Research (TIGR). Subsequently,
the genomes of more than 20 organisms have been completely sequenced by
TIGR, the Sanger Centre, Berkeley-Stanford, and other genome centers.  At
the same time, progress on sequencing the human genome is accelerating,
with completion of
the entire 3.3 billion base sequence expected soon.  This breakthrough in
genomic technology has vast implications for the life sciences and health
care.

Our research aims to develop novel approaches for learning from complex
data that help facilitate the understanding of the fundamental connections
between genetic sequence and biological function of living organisms.  We
can take advantage of the fortunate confluence of the problem of
understanding DNA and basic research questions in machine learning and
computational modeling. DNA encodes the rules of life in sequence patterns,
and many of these patterns can be deciphered using advanced computational
methods.
                                                                        
In this talk we will describe our gene identification system, Glimmer
(http://www.tigr.org/softlab). The system combines new theoretical
developments in learning with careful integration of biological knowledge
to achieve over 97% accuracy on finding genes previously annotated by
biologists. Glimmer has been already have been used to identify thousands
of genes in many whole-genome sequencing efforts at TIGR and other genome
centers. These genomes of organisms that cause Lyme Disease, Syphilis,
Tuberculosis, Malaria and many others that are currently in progress.

We will also describe a recent project for high-resolution alignment of
complete genomes. This project was supported using a new large scale
genomic comparison tool that can also be utilized for comparing human vs
mouse DNA, SNP identification in human DNA, tandem repeat finding, and
other fundamental genomics tasks.

We will conclude by discussing the important role computational genomics is
likely to play in future medical research and treatment.

This talk describes joint work with The Institute for Genomic Research
http://www.tigr.org.

Background for the talk:
"Computational Methods in Molecular Biology", eds, Steven Salzberg, David
Searls and Simon Kasif, Elsevier Publ., 2nd Ed. 1999.