Virtual Reality Display of Metabolic Networks
· Video Clips (New movies added!)
·
Links
Description
Multidimensional genome-wide gene expression profiling datasets are being generated for bacterial, plant and animal systems. This complex data reveals the interacting metabolic pathways that control cell metabolism. Ultimately, understanding these pathways will increase our ability to predict the effects of a given drug on human metabolism, the consequences of changes in a single gene on the composition of a seed, or the effect of a given mutation in a pre-cancer cell. Traditional biology research and analysis methods cannot evaluate the information in these large datasets. This project will integrate advanced biological knowledge with automated mathematical logic, complex data structures, fuzzy cognitive maps, interactive graph visualization, and other computational tools to create a novel analytical suite of tools for metabolic pathway analysis and visualization.
We will visualize and model complex metabolic pathways, developing a software system that enables scalable immersive environments, which will range from desktop visualization to the NSF-supported fully immersive C6 projection system in the Virtual Reality Applications Center (VRAC) at Iowa State. This project will lead to new methods for displaying and navigating through complex metabolic networks in three dimensions. The visual methods will be complimented by existing modeling methods that use fuzzy cognitive maps to analyze network interactions and determine their agreement with the experimental datasets.
Two experimental applications will be developed. First, a prototype interactive visualization of a metabolic network, the Calvin cycle of photosynthesis, will be used to create a living dynamic example of a biochemical pathway for K-12 students to explore. The pathway will be brought alive by a computer game that students explore at their own pace, with rewards for particular achievements such as combining the proper molecules. Complex concepts such as the conversion of chemical, light and heat energy, metabolic flux, use and synthesis of chemical constituents of living organisms can all be illustrated through this pathway. Secondly, three-dimensional visualization will be applied to the analysis of the acetyl-CoA metabolic network in the genetic model-plant Arabidopsis. This visualization will be integrated with complex datasets taken from a group of genetic mutants in specific steps in this pathway.
Navigating through a metabolic network in the C6 virtual environment
Interacting with the same application in a desktop
A snapshot of MetNetVR running
in a desktop and in the monoscopic mode. The green gadget
represents a joystick-like input device. The red line cast from the gadget
indicates the device orientation. The network contains two pathways in
Arabidopsis from the AraCyc
Database: glycerol biosynthesis and glycerol metabolism.
A user exploring a representation of gene expression levels in the C6. The plots on the left side show the clusters of genes that behave similarly. The bars through the node show the amount of variation within the cluster at each data point. The right side shows the genes that make up an individual cluster
A fan layout of reactions of interest focusing on GMP (the node in the center with a red label)
A snapshot of radial layouts of the ROIs focusing on AT5G35170 and AT3G60510; these genes have similar expression profiles. The magenta link indicates the DNA transcription of the RNA, the green indicates the protein translation by the RNA, the blue is the action of the resulting enzyme. This radial layout is exploratory and represents a set of reactions with diverse reactants (yellow lines)
A snapshot of radial layouts of the ROIs focusing on AT5G35170 and AT3G60510, without users inside
A snapshot of a mini-map used when navigating through a
large scale metabolic network. The red point in the mini-map shows the current position
of the user while navigating through the network. The network includes pieces
from 200 different pathways such as: chorismate
biosynthesis, fatty acid elongation (saturated), glucose 1-phosphate
metabolism, purine biosynthesis, isoleucine
biosynthesis I, and nucleotide metabolism.
A video clip (in QuickTime *.mov format) shows how 'AT2G22190' the putative gene for trehalose-6-phosphate phosphatase fits into the known pathways in MetNetDB. The animated portion pulled out of the main graph consists of all the reactions that gene AT2G22190 takes part in from the pathways: trehalose biosynthesis and trehalose degradation.
A video clip (*.mpeg format) (QuickTime *.mov format) shows the user navigating through the network containing two pathways in Arabidopsis from the AraCyc Database: glycerol biosynthesis and glycerol metabolism, selecting a node (Glycerol) in the scene and its detailed information being automatically highlighted in the node list a 2D GUI. The 2D GUI program is a separate program that exchanges data with the visualization program in real time. It can run either externally on a tablet PC when the visualization program running in a VR environment or in the window next to the visualization program on a desktop.
A video clip (*.mpeg format) (QuickTime *.mov format) shows the dynamic status of a biological network in a time series according to the experimental microarray data. Morphing the brightness of RNA and gene nodes as well as the transcription edges of the whole network gives the user an overall image of changes in RNA levels over the time.
A video clip (*.mpeg format) (QuickTime *.mov format) shows the dynamic behaviors of two genes
(AT3G60510 and AT5G63800) in a time series, along with the biological reactions
they take part in according to a metabolic network map.
MetNetVR-1.avi (New!) shows the
exploration of pathway-hierarchy structure of a network containing 2 pathways. The
radius of the edge connecting two pathways indicates the number of edges among
the nodes within these two pathways.
MetNetVR-2.avi
(New!) shows the exploration of location-hierarchy
structure of the same network. The radius of the edge connecting two locations
indicates the number of edges among the nodes within these two locations.
MetNetVR-3.avi
(New!) shows different methods to explore
the network
1) Using different graph
layouts (in the order of GEM3D, Weighted GEM3D and multi-level force directed
layout)
2) For each layout, adjusting
the ideal length of the edge (something like zoom-in and zoom-out, but only
applies upon edge length, has no effect on the node size and the label size.
Otherwise, labels will be difficult to see)
3) Rotation around the center
of the network to avoid lost in 3D space.
4) Moving according to
direction of the wand.
Publications
·
Dickerson
JA, Yang Y, Blom K, Reinot
A, Lie J, Cruz-Neira C, Wurtele
ES. 2003. "Using Virtual Reality to Understand Complex Metabolic
Networks." Atlantic Symposium on Computational Biology and Genomic
Information Systems and Technology, September. 950-953.
· Yuting Yang, Levent Engin, Eve Syrkin Wurtele, Carolina Cruz-Neira, Julie A. Dickerson. 2005. "Integration of metabolic networks and gene expression in virtual reality" Accepted by Bioinformatics.
· Yuting Yang, Alan Fischer, Eve Syrkin Wurtele, Carolina Cruz-Neira, Julie A. Dickerson. 2005. "Interactive Biological Network Visualization" Submitted to IEEE Visualization 2005.
Download MetNetVR (MetNet3D) for Linux:
Runtime libraries: OpenSG.tar.gz R.tar.gz
Project Personnel
- Julie A Dickerson (PI) Electrical and Computer
Engineering
- Carolina Cruz-Neira
(co-PI) Industrial Engineering
- Eve Wurtele (co-PI)
Botany
- Yuting Yang (PhD student) Computer Engineering
Funding Sources
National Science Foundation, Information Technology Research Program, Division
of Experimental and Integrative Biology. (Grant # 0219366)
Iowa State University Carver Grant for Explorative Research