2. 自己紹介
● Keiichiro ONO (大野圭一朗)
● ライフサイエンス分野に特化したソフトウェア開発者
● カリフォルニア大学サンディエゴ校医学部 Trey Ideker研究室
○ 以下の非営利団体で技術者をやっています
■ National Resource for Network Biology
■ The Cytoscape Consortium
■ Cancer Cell Map Initiative
■ San Diego Center for Systems Biology
● 主なプロジェクト
○ Cytoscape / DCell / NDEx / Cytoscape Explorler(仮)
63. Deck.gl
● Generalized framework for
data visualization
○ Just like Unity for game
developers
● Developed by data viz team
at Uber
● Open source
● Visualizes data in multiple
Layers
68. Xiaoji Chen | Feb 2020 | Contact Us
https://docs.google.com/presentation/d/1qtXUQzMuIa8NYIKUa1RKfSwvgpeccY-wrPrYqsb_8rE/edit?usp=sharing
69. Why Large Graph Renderer?
● We already have good renderer for networks on browsers, like Cytoscape.js
● It is a fully featured graph renderer
○ Has a lot of features
■ Showing charts on nodes
■ Many types of nodes/edges
○ Fully interactive
● ...But NOT designed for huge data, including millions of objects
70. Overview of the project
Main Goal: Performance
● Showing overview of millions of objects
● Has minimum set of interactivity
○ Zoom
○ Pan
○ Selection
● Just like Google Map
○ Not an editor but fast, useful viewer
○ Showing details when zooming in, display just
overview when zooming out
74. Design and Implementation
This is a React component (for web browsers)
Client Technology Stack
1. WebGL
2. Luma.gl
3. Deck.gl
4. Large Graph Renderer (LGR) ←
75. ● A React component
○ Reusable in our ecosystem
■ NDEx
■ HiView
■ Future web apps
● Fast rendering and limited interaction
○ Should handle millions of objects
○ Does not support full set of visual properties
Large Graph Renderer
76. STRING-human data set
20k nodes / 470k edges
Graph data in CX format
Auto-Graph Visualizer
https://github.com/idekerlab/auto-graph-visualizer
LGR
77. How LGR works
1. Feed the network data in CX2 data format
2. Convert the data into LGR’s format locally
3. LGR renders it