-----------------------------------------------------------------------
                       BIOINFORMATICS COLLOQUIUM
                    School of Computational Sciences
                        George Mason University
-----------------------------------------------------------------------
                    From Text Mining to Gene Mining

                              Jeff Solka
                        George Mason University

                      Tuesday, February 15, 2005
                               4:30 pm
               Verizon Auditorium, Prince William Campus


This talk will discuss recent work at applying the bipartite bipartition
methodology of (Dhillon 2001) to text and gene expression data. The
methodology has the nice property that it simultaneously clusters on
document (samples) and words (genes). Work will be discussed that
extends Dhillon's proposed methodology to perform a tree-based
clustering of documents and words. This work will be illustrated using
two small corpora (a collection of Science News articles and a
collection of Office of Naval Research Project abstracts).  Dhillon's
original methodology was also modified in order to process gene
expression data. The modified methodology was applied to the Golub
ALL/AML gene expression dataset. The ability of this technique to reveal
disease relevant genes will be discussed. The text mining work is joint
with Avory Bryant of NSWCDD and Ed Wegman of GMU. The gene expression
work is joint with Brandon Higgs of Mitre. Brandon is also a
Ph.D. student in the bioinformatics program here at GMU.

----------------------------------------------------------------------
Refreshments are served at 4:00 pm.  Find the schedule and directions at
http://www.binf.gmu.edu/colloq.html