Unboxing Cluster Heatmaps Sophie Engle Sean Whalen Alark Joshi - - PowerPoint PPT Presentation

▶
unboxing cluster heatmaps
SMART_READER_LITE
LIVE PREVIEW

Unboxing Cluster Heatmaps Sophie Engle Sean Whalen Alark Joshi - - PowerPoint PPT Presentation

Mar 15, 2023 22 likes •268 views

Unboxing Cluster Heatmaps Sophie Engle Sean Whalen Alark Joshi Katherine Pollard vgl.cs.usfca.edu docpollard.org Visualization and Graphics Lab Pollard Group University of San Francisco Gladstone Institutes, UCSF 2 Contributions

slide-1
SLIDE 1

Unboxing Cluster Heatmaps

Sophie Engle Alark Joshi vgl.cs.usfca.edu Visualization and Graphics Lab University of San Francisco Sean Whalen Katherine Pollard docpollard.org Pollard Group Gladstone Institutes, UCSF

slide-2
SLIDE 2

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Contributions

  • Practitioner Survey
  • 45 practitioners
  • Heatmap & dendrograms
  • Symmetric & asymmetric
  • 100 to 250,000 cells
  • R & Cytoscape

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

2

slide-3
SLIDE 3

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Contributions

  • Practitioner Survey
  • Unboxing Approach

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

3

slide-4
SLIDE 4

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Contributions

  • Practitioner Survey
  • Unboxing Approach
  • Pair Analytics

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

4

slide-5
SLIDE 5

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Contributions

  • Practitioner Survey
  • Unboxing Approach
  • Pair Analytics
  • Practitioner Interviews

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

5

slide-6
SLIDE 6

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Contributions

  • Practitioner Survey
  • Unboxing Approach
  • Pair Analytics
  • Practitioner Interviews
  • Large-Scale User Study

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

6

Which two elements are more closely clustered?

slide-7
SLIDE 7

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Contributions

  • Practitioner Survey
  • Unboxing Approach
  • Pair Analytics
  • Practitioner Interviews
  • Large-Scale User Study

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

7

200 5 45

slide-8
SLIDE 8

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Contributions

  • Practitioner Survey
  • Unboxing Approach
  • Pair Analytics
  • Practitioner Interviews
  • Large-Scale User Study

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

8

200 5 45

slide-9
SLIDE 9

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Unboxing Cluster Heatmaps

github.com/usfvgl/unboxing-cluster-heatmaps git.io/vw0t3

slide-10
SLIDE 10

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco 6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

10 Left: Cluster Heatmap/Gapmap, Middle: Sunburst/Circle Packing, Right: Radial Dendrogram/Force-Directed Tree

slide-11
SLIDE 11

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco 6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

11 Left: Cluster Heatmap/Gapmap, Middle: Sunburst/Circle Packing, Right: Radial Dendrogram/Force-Directed Tree

slide-12
SLIDE 12

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Design

Which two of the highlighted elements are more closely clustered?

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

12 5 by 7 circle packing

slide-13
SLIDE 13

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Design

Which two of the highlighted elements are more closely clustered?

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

13 5 by 7 radial dendrogram

slide-14
SLIDE 14

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Design

Which two of the highlighted elements are more closely clustered?

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

14 8 by 16 cluster heatmap

slide-15
SLIDE 15

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Design

Which two of the highlighted elements are more closely clustered?

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

15 8 by 16 gapmap

slide-16
SLIDE 16

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Results

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

16 Average Score Violin Plot. Circle: Average Score, Bar: Standard Error, Dotted Line: Random Performance (w/o “Unsure”)

CH-RD 0.041 GM-RD 0.040 CH-SB 0.01, CH-RD 0.02 CP-SB 0.01, CP-RD 0.02

Q13 − Close Siblings Q14 − Close Leaves Q15 − Close Clusters Q16 − Least Similar Sunburst Radial Dendrogram Gapmap Force Directed Tree Cluster Heatmap Circle Packing 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1

slide-17
SLIDE 17

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Results

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

17 Average Score Violin Plot. Circle: Average Score, Bar: Standard Error, Dotted Line: Random Performance (w/o “Unsure”)

CH-RD 0.041 GM-RD 0.040 CH-SB 0.01, CH-RD 0.02 CP-SB 0.01, CP-RD 0.02

Q13 − Close Siblings Q14 − Close Leaves Q15 − Close Clusters Q16 − Least Similar Sunburst Radial Dendrogram Gapmap Force Directed Tree Cluster Heatmap Circle Packing 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1

slide-18
SLIDE 18

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Results

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

18 Average Score Violin Plot. Circle: Average Score, Bar: Standard Error, Dotted Line: Random Performance (w/o “Unsure”)

CH-RD 0.041 GM-RD 0.040 CH-SB 0.01, CH-RD 0.02 CP-SB 0.01, CP-RD 0.02

Q13 − Close Siblings Q14 − Close Leaves Q15 − Close Clusters Q16 − Least Similar Sunburst Radial Dendrogram Gapmap Force Directed Tree Cluster Heatmap Circle Packing 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1

slide-19
SLIDE 19

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

mTurk Study Results

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

19 Average Clusters Violin Plot. Circle: Average Estimate, Bar: Standard Error

Q17 − Visual Clusters Sunburst Radial Dendrogram Gapmap Force Directed Tree Cluster Heatmap Circle Packing 2 8 14 20

slide-20
SLIDE 20

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Conclusions

  • Gapmaps are a good alternative for cluster heatmaps
  • Preferred by interviewed practitioners
  • Performed well in mTurk user study
  • Involving practitioners at multiple stages critical
  • Caught places were assumptions were incorrect
  • Symmetric matrices need (much) more study

6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

20

slide-21
SLIDE 21

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco

Unboxing Cluster Heatmaps

github.com/usfvgl/unboxing-cluster-heatmaps git.io/vw0t3

slide-22
SLIDE 22

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco 6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

22 Unboxing Approach for a Symmetric Matrix

slide-23
SLIDE 23

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco 6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

23 Mechanical Turk User Study Analysis

slide-24
SLIDE 24

Pollard Group, Gladstone Institutes, University of California San Francisco Visualization and Graphics Lab, Department of Computer Science, University of San Francisco 6th Symposium on Biological Data Visualization (BioVis), October 23, 2016 Unboxing Cluster Heatmaps by Engle, Whalen, Joshi, and Pollard

24 Mechanical Turk Post-Hoc Analysis