Transcript Gene Chasing with the Information Prism: Finding Meaningful

Gene Chasing with the
Hierarchical Clustering Explorer:
Finding Meaningful Clusters in
High Dimensional Data
Jinwook Seo and Ben Shneiderman
HCIL
Understanding Clusters
• Cluster Identification
– How to identify interesting clusters
– Interactive exploration of clustering results
• Understanding Clusters
– Expert’s domain knowledge
– Intuitive search mechanisms
– Overlap of clusters and search results
Mouse Muscle Regeneration
Time 0
12Hr
1Day
2Day
…………………………….
Measure the activities of
12,422 genes
at 27 time points
30Day
40Day
Clustering Result Overview
Time
Genes
Searching for Clusters
Minimum Similarity Bar
Searching for Clusters
Searching for Clusters
Searching for Clusters
Zoom in to
Zoom in to
Hierarchical Clustering Explorer
• 1400 downloads since April 2002
– www.cs.umd.edu/hcil/hce
– 1 commercial license to a biotech company
• Demonstration at HCIL1
• Try HCE to analyze your multidimensional data
– 3138 US Counties
– Demographic data
Sponsored by
Children’s National Medical Center