Slide 1
Slide 1 text
Feature-Centric Visual
Exploration of Genome
Interaction Maps
Fritz Lekschas, PhD candidate
Harvard University
Benjamin Bach, Peter Kerpedjiev, Michael Behrisch, Nils Gehlenborg,
and Hanspeter Pfister
July 9th, 2018
ISMB 2018
Slide 2
Slide 2 text
Cell Nucleus Contact Sequencing Heatmap
3D GENOME
Slide 3
Slide 3 text
Gene regulation:
Enhancer-promoter
Structural variation:
Break points
Cell dynamics:
Differentiation
Cell cycle
WHY CARE?
Slide 4
Slide 4 text
3 million × 3 million
Slide 5
Slide 5 text
PATTERNS
Slide 6
Slide 6 text
CHALLENGES
• Size of the heatmap
• Many called features
• No ground truth
GOALS
• Quality control
• Pattern discovery
• Pattern correlation
CHALLENGES
Slide 7
Slide 7 text
No content
Slide 8
Slide 8 text
No content
Slide 9
Slide 9 text
No content
Slide 10
Slide 10 text
No content
Slide 11
Slide 11 text
No content
Slide 12
Slide 12 text
Efficiency
Show all insets
Close to origin
Minimize occlusion
Show all insets
Slide 13
Slide 13 text
Efficiency
Show all insets
Close to origin
Minimize occlusion
Close to origin
Slide 14
Slide 14 text
Efficiency
Show all insets
Close to origin
Minimize occlusion
Minimize occlusion
Slide 15
Slide 15 text
No content
Slide 16
Slide 16 text
Performance comparison in pattern search on images and
genome interaction maps
User study: 18 participants and 6 domain experts
Techniques:
Results: faster pattern search
more accurate pattern comparison
easy-to-learn & easy-to-use
EVALUATION
Slide 17
Slide 17 text
CONCLUSION
Explore local features in
spatial context
Faster pattern search
Improved navigation
Slide 18
Slide 18 text
Thank You!
MORE INFORMATION
scalable-insets.lekschas.de
SOURCE CODE
github.com/flekschas/higlass-scalable-insets
CONTACT
[email protected]
@flekschas
lekschas.de
POSTER: B-765