CHPC News, September 1999, Volume 11, Number 1
by Zoya Maslak, Department of Art,
Douglas J. Steele, Department of Biology,
and Robert J. McDermott, Center for High Performance Computing and the Department of Computer Science, University of Utah
The molecular events involved in the activation of G protein-coupled receptors, represent a fundamental biochemical process. These events were selected for animation because the mechanism involves both a ligand-receptor conformational shape change, and an enzyme-substrate conformational shape change. This expository animation brought this biochemical process to life.
One can derive motion from different information. When the information is data containing time steps, a simulation can be depicted. The data fixes the timing for a simulation. When the information is a story based on the scientific literature, an expository animation can be created. An animator creates the timing for an expository animation. The root meaning of animation is "to give life". In our animation project we have attempted "to give life" to a foundational biochemical process: the molecular events involved in the activation of G protein-coupled receptors. The project combined the collaborative efforts of a biology post-doctoral researcher, a fine arts graduate student animator and a project director. It targeted scientific researchers in biology.
First a story was developed in consultation with a small group of local biology researchers. The story was developed to be moving, in both the physical sense, and in the intellectual sense. We wanted the story to have interesting physical movement to animate and to be a compelling story worth telling.
Once there was a solid story, a story board was drawn. The story board provided a focus for discussion with our local biology researchers about the appearance of the story to be told. Based on the story board, models could be designed. There was an extensive search for existing data to define models. Shape, color and texture of models were chosen to support the story. Model sheets were drawn characterizing the movement of the models throughout the entire story. Further discussions took place with our local biology researchers to characterize the movement of the models and to set the environment into which the models would be set.
Only after all this preliminary work, was time devoted to creating models for the animation and computer animation software for the movement. SoftImage 3D Animation Software was chosen for its balance between complexity of modeling and ease of use by an animator. Models, as well as an environment for these models, were produced using SoftImage 3D.
The story was broken down into a series of scenes. Subsequently, line drawing of the animation, followed by 1/2 resolution double-time flip books of the animation, and finally, 30 frames-per-second video resolution rendered images were produced for each of the scenes. As each scene unfolded, many discussions took place about the appearance and movement with our local biology researchers.
After approximately a year's period of time and many hours of reviewing animation, we produced 8 scenes consisting of some 7,000 final rendered images. The images were transferred to a SoftImage Digital Studio for editing, where voice over and music were added. From a digital video master, analog video duplicates have been produced for viewing in this year's video proceedings as a companion to this case study paper.
A schematic for G protein-coupled receptor activation (Fig. 1) is currently used in textbooks . Initially, neither the ligand molecule nor the G protein complex is associated with the receptor (Fig. 1a). Once the ligand binds to the outside of the receptor (Fig. 1b), the G protein complex attaches to the inside of the receptor. A molecule of GDP detaches from the receptor-G protein complex and is replaced by a molecule of GTP (Fig. 1c). This allows the G protein complex to come off the receptor and dissociate into it's respective subunits (Fig. 1d). The alpha subunit/GTP diffuses to a molecule of an enzyme adenylate cyclase, binds to it, activates it, and initiates a cascade of cAMP production (Fig. 1e). Eventually, the GTP is hydrolyzed to GDP, and as a result the alpha subunit/GDP falls off the adenylate cyclase and re-associates with the other G protein subunits (Fig. 1f). The ligand dissociates from the receptor and the system returns to its starting conditions, awaiting re-activation (Fig. 1g).
The brain is filled with millions of neurons  specialized for sending and receiving electrical signals. They have two types of long finger-like projections which form billions of connections with other cells in the brain, to communicate instructions for the body. The axons are slender extensions which terminate in a flattened bulb called a bouton. It is from the boutons that chemicals are released to communicate with other cells. The second projections are called dendrites, which are dense tree-like networks of nerve endings which make contact with the axons and receive the chemicals released from the axons. It is through a combination of electrical impulses and the release of chemicals that messages are relayed throughout the brain.
Line drawings of the hypothesized three-dimensional structure of a G protein-coupled receptor have been derived from data . Starting with the amino acid sequence of a receptor, this model has been constructed to take into account secondary and tertiary structural features of the receptor including alpha-helical and beta-sheet regions, hydrophilic and hydrophobic domains, the seven trans-membrane domains, as well as intracellular cytoplasmic loops and extracellular receptor-binding regions. Until the complete X-ray crystal or NMR structure of such a receptor is solved, such diagrams represent the best overall model of G protein receptor structure.
There were design problems to be solved for models, colors and animation. The models were designed to capture both an appearance and a movement, colors were used for identity and activity of the models, animation was used to depict both interaction between models as well as activity within a model.
Models made use of techniques that supported both overall shape, and the internal movement of a model. The neurons were created by using of B-spline tubes (Figure 2).The B-spline tubes were chosen so that the animated models strongly resembled the image of physical neurons. Because of the dynamic nature of the bio-chemical process of the G-protein Coupled Receptor and Adenylate Cyclase enzyme activation, the best tools we found were Meta -Clay modeling and Meta-Clay shape animation (Figure 3), To save rendering time, models of the receptors in the background were converted from Meta-Clay models to polygon models.
Because the range of colors available for video is considerably less than the range of colors for painting or film, the colors for models were chosen from a constrained range. Shading of a single hue was used to depict activity of a model. Low intensity, quiet colors were chosen for the inactive receptors (Figure 3). Bright, high intensity colors were chosen for the active receptors. 3D Textures were used to depict both shape change and activity. A 3D texture was applied to the model of simple static polygonal sphere representing the nucleus of the neuron model.
The main concern of our animator was to create a believable consistent movement through the medium, combined with clear and easy-to-understand changes of shapes. The environment depicted objects immersed in a liquid-like substance. Camera fly-throughs were performed to move the viewers point of view through out the environment without any visible "collisions" with objects. The complexity of the nucleus was depicted by the choice of an animated 3-D texture. To depict the activation and deactivation of the receptor and enzyme, a combination of shape-animation and color animation were used. Meta-Clay shape animation allowed for complex yet subtle changes within the G protein receptor model and more dramatic changes within the model of the Adenylate cyclase enzyme.
Video production creates a "package" to contain information and "frames" results in a way that it is easily understood. From the start of our project we planned on delivering our animation in video. An archival form, as well as a final product can be repeatedly viewed and easily distributed.
The visual editing was straight forward due to the preplanning of our animation. During the visual editing some animation errors were seen and they were corrected. These errors were not seen prior to the visual editing. Previously, we viewed flip books of the animation by repeating frames two or three times, this was due to the amount of time needed to render these flip books. At the time of the visual editing we were viewing all of the full frame images for the first time, hence some errors.
The voice-over attempted to achieve a delicate balance between images and words. A script was developed while viewing the animation of each scene played over and over in a loop. There were the traditional clashes between script writer, our biology post-doc, and the director. The script writer kept asking to stretch the final animation in order to fit more words. In the end the timing of the animation won out and the words were reduced.
The many hours of work developing our script allowed us to record the entire voice-over in one take and fit it directly to the entire animation. This was a brilliant stroke of good fortune.
Initially, music was composed in three sections by Uri, a friend of Zoya's, who was only looking at the story board. The first section was light and fit much of the animation. The second section was somewhat more dramatic and appropriate for the more active portions of the animation. The third section was very dramatic and may have sent our viewers home in tears, so this music was not used in this animation project.
The final music was composed by Uri, while viewing a draft edit of the full animation on video tape. This composed music was appropriate in both pace and mood for our animation. The music fit exactly in time and was recorded in one pass of the full animation. This was another brilliant stroke of good fortune.
An abstract and a video tape of our animation were submitted to the 17th International Meeting of "Molecular Modeling in the Large", organized by Art Olsen, Director of the Molecular Graphics Laboratory of Scripps Research Institute. Art has been producing computer graphics and computer animation of biological processes for many years. Our submissions were accepted  and put in a session where there was an invited paper entitled, "Visualization of the Receptor-G-Protein Interface" by Garland Marshall, Center for Molecular Design, Washington University. This invited speaker's title seemed as if it related directly to the content of our animation. Only after we had heard the invited presentation delivered were we sure that our submission would indeed be a unique contribution to the meeting.
We delivered our presentation and showed a video tape of our animation which were both very well received by the attendees. Most attendees felt that the events portrayed were an accurate representation of the scientific literature. The only point raising scientific controversy was the loose tethering of the complex attached to the membrane. On this point, there was a division of opinion, which usually characterizes a point that is not fully understood.
Two still images from our animation (Figures 2 & 3) were rendered at 3000 lines and plotted at 22" by 28". These images were framed and submitted to the 1998 Molecular Graphics Art Show curated by David Goodsell of Scripps Research Institute . Both stills were accepted for the exhibit and were well received at the opening of the exhibit by the attendees and other contributing artists.
Teresa Larsen, a member of the Molecular Graphics Laboratory of Scripps Research Institute, who produced an expository animation entitled "Looking Into HIV", juried the exhibit  and attended the meeting. She entered into lengthy discussions about the experience we have gained during our project relating to the experience she has accumulated over a number of years. An in depth exchange took place between individuals aspiring to the same lofty goal of portraying scientific stories with animation.
Fred Brooks of the University of North Carolina, who has worked for many years on visualizing biological processes, also attended the meeting. He was very encouraging and instigated our giving a lecture to the National Science Foundation, Science and Technology Center, Graphics & Visualization Center's graduate graphics course, which is offered to the five affiliated universities; Brown University, Cornell University, University of North Carolina, University of Utah and California Institute of Technology. It was gratifying that our project had something to offer this community of computer graphics researchers.
We have also presented this work as part of a Center for High Performance Computing seminar series for researchers in our Intermountain Networking and Scientific Computing Center at the University of Utah. This community consists of researchers with computationally intensive work in Chemistry, Physics, Mathematical Biology, Meteorology, Geology & Geophysics, and Chemical & Fuels Engineering. This community questioned our use of music and our seemingly continuous portrayal of time. We responded that the music was used to help both the flow of the animation and to provide an underlying supporting mood for the animation. As to the question of timing, we brought to their attention that within a scene there is the portrayal of an event but that between each scene there may well be additional time due to the nature of the Brownian motion of the objects.
An animation project benefits from many contributions, such as people, moving story, story board, model sheets, modeling, wire frame animation, flip-book animation, rendering, frame storage & transfer, visual edit, script, voice over, music and audio edit. Considerable effort was made to achieve a balance in the many contributions so that the finished production would possess an over-all high quality.
The goal of this animation was to provide researchers with a view of biochemical processes occurring in living cells. Scientists have images of complex biological processes in their imaginations, but those images are very different from one to the next. The feedback from researchers who have viewed the animation has been positive with respect to providing an animated visualization for this particular biological process.
The resulting animation, we feel, was well worth the individual efforts, as well as the expenditure of computational resources.
Doug Steel and Toto Olivera's research group in our Department of Biology provided endless hours of critical discussions and viewing of our work throughout the entire project. They provided our project with the critical scientific input needed for an animation project like this to succeed.
Zoya Maslak endlessly persisted as a designer-animator over the entire year. Her willingness to respond to the scientific criticism, as well as the criticism of all aspects of the animation, truly tested both her Russian up-bringing and early education.
Computational support, animation software, digital video and audio editing for this project were provided by the Center for High Performance Computing at the University of Utah.
-  Traditional Representation of G Protein-Coupled Receptor Activation from: Schwartz, J.H. and E.R. Kandel, Principles of Neural Science, Third Edition, Elsevier, New York, 1991, p.176.
-  Cellular Architecture of a Neuronal Network from: Hinton, G.E., How Neural Networks Learn from Experience, Scientific American, 267(3):144, 1992.
-  Swiss Protein Databank, Robert Bywater Novo Nordisk, A/S DK-2880 BAGSVAERD, Denmark email: firstname.lastname@example.org http://www.rcsb.org/pdb
-  Z. Maslak, D.J. Steel, R.J. McDermott, A computer animation representing the molecular events of G protein-coupled receptor activation, Molecular Graphics And Modelling Society 17th International Meeting, San Diego, CA, 6-10 December 1998,. http://www.mgmsoa.org
-  1998 Molecular Graphics Art Show, Reuben H. Fleet Space Theatre and Science Center, Balboa Park, San Diego CA, December 2,1998 - January 15, 1999. http://www.scripps.edu/pub/goodsell/mgs_art/msg_arts/index.html
-  Teresa A. Larsen, 1998 Molecular Graphics Art Show, Juror's Comments. http://www.scripps.edu/pub/goodsell/mgs_art/msg_arts/ comments2.html.
by Julia Harrison, Assistant Director, User Services, CHPC
The SP at CHPC recently underwent a major upgrade. The operating system was upgraded to AIX 4.3.2, the batch system changed from LoadLeveler to PBS and some additional resources were added (disk space and 10 new nodes).
The upgraded SP installation consists of: 74 Nodes:
- 8 - 66 Mhz thin nodes: POWER2 Processors. 5 - 128 MB memory nodes used for interactive and development and 3 (2 128 Mb, 1 256 Mb) used for system utilities.
- 56 - 120 Mhz thin nodes: POWER2 SC Processors. 4 nodes - 1 Gbytes of RAM, 2 nodes - 512 Mbytes of RAM, 2 nodes - 256 Mbytes of RAM, and 48 nodes - 128 Mbytes of RAM.
- 8 - 160 Mhz thin nodes: POWER2 SC Processors. 1 Gigbyte memory each
- 2 Nodes 332 Mhz each - 2 way SMP (4 processors). 2 Power3_SMP Processors each node. 1 Gigbyte memory each.
Other features include a global shared filesystem, full version of AIX 4.3 each node, the networking changed from FDDI to atm, ssh is now available and highly recommended and the high performance switch.
Users can access the SP interactive nodes via ssh or telnet to sp.chpc.utah.edu. There are 2 interactive nodes and connections will be established in a round robin fashion. This number may be changed if the load requires it.
New on our SP is a large scratch filesystem globally available to all nodes. There is approximately 100 Gbytes available. The path to this disk space is /global/sp/.
We recently installed gaussian98 on the upgraded SP. There is a change in how you indicate the number of processes you want to use. On the SP only you need to specify %NPROCL (on all our other platforms you use %NPROC).
Please let us know if you have questions or problems using the updated system. (email to email@example.com or call 581-4439).
by Julio Facelli, Director, CHPC
The University of Utah has been connected to the vBNS backbone since March of 1998; this connectivity has greatly enhanced the ability of our researchers to communicate with other universities, supercomputer centers and research institutions. The vBNS connection was partially funded by a NSF grant that will expire in March of 2000. Working in partnership with Utah State University and the Utah Educational Network the University of Utah will shift its high performance network traffic to Abilene.
Abilene (http://www.ucaid.edu/abilene/html/participation.html) is a project of the University Corporation for Advanced Internet Development (UCAID) together with several major telecommunications corporations. It is being organized by UCAID in support of all its members, including participants in the Internet2 project.
The University is in the process of connecting to Abilene at OC3 speed, the same as our current vBNS connection. We expect to begin passing traffic on the Abilene backbone by early October. The vBNS connection will remain operational until March 2000 and we do not expect any disruption of services or performance degradation as we migrate from the vBNS to the Abilene backbone. Funds have been identified to secure this level of advanced network services until July 2001. We will continue monitoring the utilization of these links to determine when an upgrade to higher speeds may be necessary.
Through our continuous participation in advanced network forums like Internet2, and consultation with relevant users, we will develop a plan to continue providing advanced network services beyond July 2001.
If you have any questions or comments on our vBNS and/or Abilene connections please contact Julio Facelli (firstname.lastname@example.org) or John Storm (email@example.com).
The Center for High Performance Computing is organizing an exhibit at SC99, the largest convention on high performance computing, to be held at the Oregon Convention Center, Portland, Oregon, November 13-19, 1999.
Research exhibits at SC99 provide an opportunity to demonstrate new and innovative research results.
The CHPC is planning to exhibit relevant research in high performance computing and visualization at the University of Utah.
CHPC is soliciting research groups to participate to this exhibit and to contribute to its success. If your group is involved in high performance computing and visualization and you have appropriate material, please contact Julia Harrison (firstname.lastname@example.org or 581-5172), or Robert McDermott (email@example.com) who are coordinating the research exhibit.
More information on SC99 and research exhibits can be found at http://www.sc99.org/.
by Brian D. Haymore, Systems Administrator
Last December the CHPC deployed a Linux Beowulf Cluster to test and evaluate this newly popular super computing design. It consisted of 32 single processor Intel Pentium II 350 computers with 256MB of RAM each and a development node that housed 60GB of disk space for user applications.The machine was named ICE Box for Intel Cluster Experiment as that is what is was, an experiment. ICE Box was put into production and we began to learn and understand the strengths and weaknesses of our design.
Overall we were very pleased with the results we saw. Because of the success of the system the decision was made by CHPC and the Henry Eyring Center for Theoretical Chemistry to further invest in this system and expand it and try to polish off some of the rough spots in our first design.
We started by buying a dedicated switch for ICE Box. We purchased a Cisco 6009 switch which yields a 35 gigabit backplane. This makes it able to handle up to 350 compute nodes before we over subscribe it's capacity. It also has two gigabit ethernet ports.
We then redesigned the development node to split up it functionality. We created three nodes to do the work the original one did. First we have an interactive development node for the users to log into, develop and deploy their applications. Second, we have an administrative node responsible for all account information as well as the scheduling of jobs. Third and last, we have a dedicated file server using gigabit ethernet and hardware raid to house 160GB of user disk space. We also reworked the shelving design to allow for better access to the machines for servicing.
The specifications for the current design of ICE Box are as follows: 22.5 GB of RAM, 84 Intel Pentium II 350 processors, 295 GB of scratch space (3.5 GB per node), 160 GB of user disk space.
ICE Box is not generally available however, if you feel you would like to try your application on this new architecture, please request it by sending email to our director, Julio Facelli, (firstname.lastname@example.org).