|
Activity Number:
|
324
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statisticians in Defense and National Security
|
| Abstract - #306526 |
|
Title:
|
Anomaly Detection in Genetic Networks
|
|
Author(s):
|
Christopher Overall*+ and Jeffrey L. Solka and J. W. Weller and Carey Priebe
|
|
Companies:
|
George Mason University and Naval Surface Warfare Center and George Mason University and Johns Hopkins University
|
|
Address:
|
5316 Satterfield Drive, Woodbridge, VA, 22193,
|
|
Keywords:
|
scan statistics ; genetic networks ; microarrays ; anomaly detection ; chatter
|
|
Abstract:
|
This talk will detail our recent work on the application of graph-based scan statistics to genetic networks in order to detect anomalies in gene activity. An anomaly may manifest as a single gene with excessive connections to other genes or it may manifest as "chatter," in which genes centered about a particular gene exhibit excessive connections with one another. This work has been adapted from previous efforts by Priebe et al. (2005) to detect anomalies within the social network developed from the Enron email dataset. Our methodology uses multivariate model-based clustering to create a time sequence of graphs from time-series gene expression datasets. It then applies the graph-based scan statistics methodology to detect anomalies within the graphs for each time point. The procedure will be illustrated using a well-known Drosophila gene expression dataset.
|
