Overview of presentations and posters at VIZBI 2013 - Visualizing Biological Data http://vizbi.org/2013/Program/

1. VIZBI 2013
March 20-22
The Broad Institute of MIT and Harvard

2. “How Can We Fix The Display?”
Visual Analytics and HCI
 Visual variables
 Colour, size, motion, luminance, shape etc.
 Change detection
 Size > Colour > Orientation
 How much realism?
 Too much requires inference
 Animation
 New - longer knowledge of static design
 Back button
 Speed control – many animations too fast
 Sweet spot of complexity vs inference
 Controls to add/remove information
 Training / domain knowledge important
 Good visualization requires shorter training
 Studies – eye tracking, emotional response

3. Visual Design Principles
 Consistency, concise, no redundancy, clarity,
focus/emphasis, truth/accuracy/detail
 Can’t compare network layouts side by side
 Optical illusions – problem for heat maps
 High contrast for legibility
 Visual weight
 Don’t give different things same visual weight
 Use gradient of visual weights instead of colour to
show change e.g not obvious purple leads to red
 Then use colour to delineate e.g. different cell
populations using same weighting scheme

4. Communicating Science Visually
 A representation, not the thing
 Photo, visualization etc.
 Clarifies the work (to ourselves and others)
 Striking photo or model
 Improve lighting, angle etc
 Can make look too perfect
 Metaphor
 e.g. pin art to illustrate rastering
 Avoid factual errors
 Hand-drawing: a representation, a process
 Poster design
 Panels in groups
 More space
 Delineate text – boxes, shadows
 “Visual abstract”
 See large images from far away

5. Designing for Different User
Groups
 Software designer’s idea of requirements
is different from user’s
 Make requirements gathering part of
research
 User registration – can group and store

6. Epigenetics
 Single genotype -> multiple phenotypes
 Malignant tumour analysis with IGV/cBio Portal
 Network graphs from transcription factors
 Red/blue edges: pos/neg regulation
 WASHU Epigenome Browser
http://epigenomegateway.wustl.edu
 Genome/metadata heatmaps
 Genome3D
 http://genomebioinfo.musc.edu/Genome3D/
 Structural, genomic, epigenetic data viewer

7. IGV/cBio Network Viewer

8. Caleydo / enRoute
 enRoute – extract pathway data from
KEGG / WikiPathways
 Display pathway vertically
 Display experimental data horizontally in
groups
 http://www.icg.tugraz.at/project/caleydo
/projects-1/enroute

9. Caleydo / enRoute

10. Visualizing RNA
 http://www.slideshare.net/ppgardne/vizbi
2013-visualising-rna
 Rfam database
 Sister to Pfam (protein)
 Aim to annotate all ncRNA families
 Analysis of C/D box snoRNA taxonomy
 Sunbursts
 Concentric pie charts
 External ring contains child nodes of internal ring

11. Sunburst

12. Visualizing Transcript Data
 Non-coding RNA
 Experimental/computational variation
 Bowtie – parameters -> different results
 RNA-Seq
 Transcriptome deep dequencing, levels, isoforms
 Tuxedo suite of sequencing tools
 Cufflinks/Cuffdiff2
 Isoform resolution and splice variants
 CummeRbund
 Plots + heatmaps

13. Beyond the Hairball
 Network graphs don’t scale up - become “hairballs”
 Improvements
 Clustering eg Cytoscape + clusterMaker
 Collapsing nodes
 Statistical analysis before visualisation
 3D
 Alternatives
 BioTapestry
 Create submodels and drill down
 BioFabric
 Nodes are horizontal lines
 Edges are vertical lines

14. BioTapestry

15. BioFabric

16. Structures & Features
 Protein structures
 Crystallography
 NMR
 “Sequence-Structure Gap”
 Cost per genome sequence has fallen
 Structure resolution difficult and expensive
 SRS 3D – integrate sequence, structure, gene features
 Aquaria WS (late 2013)
 Scale to large screens
 Hardware acceleration
 Augmented reality navigation
 Structure viewer – Java applet – to be redeveloped in WebGL
 Sequence viewer – D3.js
 Cross-highlighting between sequence and structure

17. Aquaria WS

18. Comparison and Assemblies
 UCSF Chimera
 Molecular structure visualization
 Density maps, sequence alignment, docking
 Animation – create storyboard like iMovie
 Linear interpolation morphing between frames for smooth animation
 Autopack – packing algorithm
 Different types of transition between frames
 Rock, rotate, morph etc.
 Interaction
 3D glasses
 Motion detection – hand gestures
 Trackpad gestures
 Navigation – can clip through panes of the structure
 WebGL export of animation storyboard
 structureViz Cytoscape plugin

19. Evolution of Protein Structure
and Function
 Protein superfamiles
 Separated by billions of years
 No sequence similarity
 Structure conservation
 FunTree
 Annotation of evolutionary branches
 ITOL
 Circular graph with tree of life at centre
 CATH
 Structure classification
 Genome3D
 Predict structural domains from protein sequence
 Future
 Develop library of JavaScript/HTML5 components using D3.js

20. Biological Networks
 Network Biology
 Emerging field
 Elements of systems biology, bioinformatics etc.
 Nature Paper: Network biology: understanding the cell's
functional organization. (Barabási + Oltvai 2004)
 Networks an anchor for other visualizations
 Easy for biologist to understand
 Pathway – a type of network
 Similarity networks
 E.g protein-protein, chemical-chemical, co-expression

21. Biological Networks: Opportunities
 Pathway automatic layout
 Hairball – cluster to provide structure and colour
 Clustering – mostly partitioning
 Fuzzy, time-variant
 Network comparisons
 Between states or species
 Over time e.g. post-translational modification
 Connect structure information to nodes
 Large data
 Progressive disclosure
 Collapsing
 Stop reinventing
 Integrate with existing platforms’ plugin architectures
 Google Summer of Code
 BioFabric – edges just as important as nodes

22. NIH LINCS Project
 http://www.lincsproject.org
 Cellularresponse to perturbation
 Catalogue changes in gene expression
and cellular processes
 Cmap Data Explorer
 Query up and down-regulated genes
 Search for signatures

23. Physiology and Function
 3D Slicer
 Build up 3D model from 2D images
 Load files on the fly so whole model not in memory
 WebGL export – share with collaborators
 Future – WebCL GPU computing, augmented
reality interaction
 http://goxtk.com/
 WebGL library for scientific visualization
 Entire presentation in WebGL!
http://danielhaehn.com/p/vizbi2013/

24. Others
 Developmental Anatomy
 GoFigure – time-lapse
microscopyhttps://wiki.med.harvard.edu/SysBio/Megaso
n/GoFigure
 Genes and Geometry
 3D pheotyping http://www.mouseimaging.ca/
 Average voxels to find true homologous points
 Supramap http://supramap.org
 Integrate genetic/geospatial/temporal data e.g. spread
of SARS
 Metagenomics
 QIIME microbial community analysis http://qiime.org
 Kbase – predictive systems biology

25. Posters
 BioTapestry – nodes are lines
 Shiny – Rstudio to web application
 Chimera – stucture modelling & animation
 Bombastic – clustering gene expression
 WASHU epigenome browser
 Visualizing molecules
 Metalwork sculptures
 WebGL application

26. Posters
 VIZBI Plus
 Public engagement in Australia
 Working with animators e.g. The Hungry
Microbiome – intestinal flora animation
 OME/OMERO – Image data management
 BioVis contest
 Predicting impact of mutations on proteins
 http://biovis.net/contest/

27. Posters
 Connectivity Map – transcriptomics - LINCS
genome-wide transcriptional expression data
 Clickme – Generate HTML from R
 3D modelling of Streptomyces growth in WebGL
 OneZoom – fractal tree of life explorer
 Interactive Visualization of Biomolecular
Simulations – GPU ray casting
 Aquaria – protein structure web app
 TRNDiff – multiple transcriptional regulatory
networks visualized in D3.js

28. Posters
 InVeo
 Network models of layers and connections
 Layers for genome, transcriptome, proteome,
metabolome
 CBioPortal.org – cancer genomics gateway
 Cell Signaling Pathways
 HTML5 iPad app (ProMega)
 Streamgraphs
 Temporal changes in gene expression in marine
microbial communities

29. BioLayout Express3D Poster

30. BioLayout Feedback
 Alternative layout algorithms
 Navigation – gestures
 Plugin architecture – DB access, file
formats etc.
 Pathway animation – input experimental
values rather than simulation
 Align multiple networks
 3D protein structures on nodes
 Pre-clustering