Multi-View Design Patterns & Responsive Visualization for Genomics Data
1. Multi-View Design Patterns &
Responsive Visualization for
Genomics Data
Sehi L'Yi Nils Gehlenborg
Biomedical Informatics, Harvard Medical School
@sehi_lyi @ngehlenborg
4. In many hospitals, screens are very old, and this often relates
to the amount of information that we can show. ... [The name
of a front-end engineer] always thinks of how this interface
looks like on small screens.
— Project Manager of Interactive Data Portal1
1 CGAP (Clinical Genome Analysis Platform): https://cgap.hms.harvard.edu/
“
5. In many hospitals, screens are very old, and this often relates
to the amount of information that we can show. ... [The name
of a front-end engineer] always thinks of how this interface
looks like on small screens.
— Project Manager of Interactive Data Portal1
1 CGAP (Clinical Genome Analysis Platform): https://cgap.hms.harvard.edu/
“
How To Support Smaller Screens In Genomics Data Visualizations?
Responsive Multi-View
How To Support Smaller Screens In Genomics Data Visualizations?
Vertically Long
A-VERTICAL Cross-Shaped
D-CROSS
Horizontally Wide
B-HORIZONTAL
Circular
C-CIRCULAR
Vertically Long
A-VERTICAL
B
Circular
C-CIRCULAR
Vertically Long
A-VERTICAL Cross-Shaped
D-CROSS
Circular
C-CIRCULAR
Cross-Shaped
D-CROSS
Horizontally Wide
B-HORIZONTAL
9. Survey Design
• 40 out of 188 Tools1-2
• Web-based genome-mapped data visualization
• 5 Logical Resolutions (3 Devices x 2 Mobile Orientations, Chrome DevTools)
• 1920×1080 (Desktop)
• 1024×768 (Google Nexus Tablet)
• 375×812 (iPhone X)
9
1 GenoCAT (http://genocat.tools/)
2 awesome-genome-visualization (https://cmdcolin.github.io/awesome-genome-visualization/)
10. Key Insights
• Genomics Data Visualization Uses MANY Views
• Average 8.1 views (SD=9.4) by default
• Majority (52.5%) allowed adding views
10
(Chen et al. TVCG 2021) "Less than five views"
→ Screen scalability challenges responsive designs
11. Key Insights
• Genomics Data Visualization Uses MANY Views
• The Four Composition Patterns Are Most Typical (92.5%)
11
Vertical Horizontal Circular Cross-shaped
12. Key Insights
• Genomics Data Visualization Uses MANY Views
• The Four Composition Patterns Are Most Typical (92.5%)
12
Vertical Horizontal Circular Cross-shaped
(70%) Horizontal juxtaposition was most popular (Chen et al. TVCG 2021)
13. Key Insights
• Genomics Data Visualization Uses MANY Views
• The Four Composition Patterns Are Most Typical (92.5%)
• Responsive Designs Are Limited In Genomics Tools
13
Responsive Design Taxonomy (Kim, Moritz & Hullman, CGF 2021)
14. Key Insights
• Genomics Data Visualization Uses MANY Views
• The Four Composition Patterns Are Most Typical (92.5%)
• Responsive Designs Are Limited In Genomics Tools
14
Usability Categories (Wu et al. TVCG 2020)
View Out
View Group Out
Visualization Out
Unreadable
Occlusion
The Number of Usability Issues Found
15. Key Insights
• Genomics Data Visualization Uses MANY Views
• The Four Composition Patterns Are Most Typical (92.5%)
• Responsive Designs Are Limited In Genomics Tools
15
Usability Categories (Wu et al. TVCG 2020)
View Out
View Group Out
Visualization Out
Unreadable
Occlusion
The Number of Usability Issues Found
Examples Reproduced Using Gosling
Readability Issue View/Visualization Out of Viewport
20. Responsive Multi-View Designs
20
Change Arrangement
View Composition Taxonomy
(Kim et al., Gleicher et al., L'Yi et al., ...)
Complementary Matrix
Liu and Shen, CHI 2015
21. Responsive Multi-View Designs
21
Change Arrangement
View Composition Taxonomy
(Kim et al., Gleicher et al., L'Yi et al., ...)
Change Layout
(Nusrat et al., Kim et al.)
22. Responsive Multi-View Designs
22
Change Arrangement
View Composition Taxonomy
(Kim et al., Gleicher et al., L'Yi et al., ...)
Change Layout
(Nusrat et al., Kim et al.)
Shorten Distance
(L'Yi et al.)
Overlay Guideline
(L'Yi et al.)
23. Responsive Multi-View Designs
23
Change Arrangement
View Composition Taxonomy
(Kim et al., Gleicher et al., L'Yi et al., ...)
Change Layout
(Nusrat et al., Kim et al.)
Shorten Distance
(L'Yi et al.)
Overlay Guideline
(L'Yi et al.)
29. Takeaway
29
Responsive Designs Needed
In Genomics Visualization
4 Composition Patterns
8.1 Views On Average
Screen Scalability Issues
Survey
Approaches
Change Arrangement Change Layout Filter
Shorten Distance
Overlay Guideline
Grammar
... Concept Is Intuitive
👍 Dependencies to Gosling
👎
30. Multi-View Design Patterns &
Responsive Visualization for
Genomics Data
Sehi L'Yi Nils Gehlenborg
Biomedical Informatics, Harvard Medical School
@sehi_lyi @ngehlenborg
https://gosling.js.org/