1. David Abramson & Hoang Anh Nguyen Monash University

2. Background ◦ Scientific Workflow ◦ Tiled Display Wall ◦ Why do we need a SWF-TDW link ? Design and Implementation Case Study Conclusions & Future works

3.  In-silico science (e-Science) ◦ Complex process ◦ Multiple steps in different computing environment  Scientific workflows ◦ Help automate, manage and execute steps ◦ Provide a high level, robust, repeatable research environment.

4.  SWF technology ◦ Application of workflow technology to solve scientific problems [1] ◦ Different from Business Workflow  SWF Management System (SWFMS) ◦ Specification, modification, run, re-run, and monitoring of workflows  Number of SWFMSs: Kepler, Taverna, Triana, Vistrails, etc.  Kepler was chosen to implement our prototype

5.  Built on top of Ptolemy II ◦ Actor-oriented modelling ◦ Vergil user-interface  Actor-oriented ◦ Actors with input/output ports ◦ Director  Powerful SWFMS ◦ Web and grid-services support ◦ Provenance information

6. Figure 1: Sample Workflow in Kepler (source: [2]) ‏

7.  What is a TDW ? ◦ Visualization cluster ◦ Multiple displays controlled by a powerful computer/cluster ◦ Acts like one or many virtual displays  TDW could be ◦ Projectors ◦ LCDs

8. Figure 2: Scalable Display Wall view from the back (Source [3]) ‏

9. Figure 3: An Optiportal at Monash Clayton ( Source [4] ) ‏

10.  Built on top of Rocks  Using SAGE, CGLX, COVISE as rendering middleware  SAGE: Scalable Adaptive Graphics Environment ◦ Open source ◦ Distributed architecture: decouple graphic rendering and graphic display

11. Figure 4: SAGE architecture SAIL: Sage Application Interface Library Sage receive r Sage receive r Sage receive r Sage receive r Sage receive r Sage receive r Free Space Manage r Free Space Manage r UI Clie nt SAIL App 1 SAIL App 2 App 3 SAIL SAGE messages Pixel stream

12.  Natural marriage ◦ Computation and visualization  To date, no easy method connecting SWF to TDW. ◦ Manual process ◦ Did not receive a lot of attention from workflow community

13.  Goals: ◦ Provide seamless link between SWFs and TDW ◦ Middleware independence ◦ Future user interactions  Design Alternatives ◦ SSH actor ◦ SAGE actor ◦ Distributed architecture: dedicated server

14. SSH protocol  Simple  Inflexible  Simple  Inflexible SSH Actor Figure 5: Solution using SSH actor

15. messages Pixel stream Sage receiver Sage receiver Sage receiver Sage receiver Sage receiver Sage receiver Free Space Manag er Free Space Manag er UI Client App SAIL SAGE actor UI Client JNI Kepler code (Java) SAIL Figure 6: SAGE actor block diagram  compact  possible feeding user feedbacks to workflow  intensive computation on machine running Kepler  middleware dependent  compact  possible feeding user feedbacks to workflow  intensive computation on machine running Kepler  middleware dependent

16. Figure 7: Distributed Architecture ‏ Server Interface OptIPortal Middleware OptIPortal Middleware Server Interface OptiServer OptIPortal Kepler OptIPortal Actor OptIPortal Middleware OptIPortal Middleware OptIPortal Middleware OptIPortal Middleware  middleware- independent  highly distributed  small communication overhead  middleware- independent  highly distributed  small communication overhead

17. Figure 8: Implementation messages Pixel stream Sage receiver Sage receiver Sage receiver Sage receiver Sage receiver Sage receiver Free Space Manag er Free Space Manag er SAIL App 1 SAIL App 2 App 3 SAIL OptiUI Client OptiServer Kepler OptIPortal Actor

18.  OptiportalActor ◦ Stream files ◦ Communicate with OptiServer  OptiServer ◦ Background process controlling the visualization  OptiUI ◦ Custom SAGE UI

19.  Illustrate the ease of use with OptiportalActor  Use OptiportalActor in a set of optical microscopy workflows ◦ To visualize images of antibody cancer therapies*  Part of a larger project ◦ Virtual microscopy ◦ Demonstrating the utility of workflows for microscopy

20.  Developed in the Faculty of Medicine, Monash University  Fluorescent labeled antibodies, together with various reagents, are used to mark three distinct tissue types: ◦ tumour nuclei ◦ “stroma” or connective tissue ◦ blood vessels that feed the tumour  These therapies work by denaturing the blood vessels to the tumor

21. Figure 9: Cancer Nuclei, Blood vessels, Stroma in confocal microscopy Nuclei Stroma Blood vessels Merged image

22. Figure 10: Confocal scanning workflow

23. Figure 11: Cancer image stack on Optiportal

24. Figure 12: Therapy effectiveness measurement workflow

25. Figure 13: Therapy effectiveness calculation on Optiportal

26.  SWF-TDW linkage  Demonstration the system with a case study in optical microscopy  To-dos ◦ Support more data-types (currently images and movies) ◦ Support other middleware ◦ Support more interactive modes of operation: computational steering environment.

28.  [1] Lin, C., Lu, S., Lai, Z., Chebotko, A., Fei, X., Hua, J. and Farsha, F. “Service-oriented architecture for view: A visual scientific workflow management system.”, In SCC ’08: Proceedings of the 2008 IEEE International Conference on Services Computing, pages 335–342, Washington, DC, USA, 2008. IEEE Computer Society.  [2] https://kepler-project.org/users/copy_of_LotkaWorkflow.png/image_largehttps://kepler-project.org/users/copy_of_LotkaWorkflow.png/image_large  [3] http://systems.cs.princeton.edu/omnimedia/images/back24.jpghttp://systems.cs.princeton.edu/omnimedia/images/back24.jpg  [4] http://messagelab.monash.edu.au/Infrastructure/OptiPortalhttp://messagelab.monash.edu.au/Infrastructure/OptiPortal  [5] http://www.sagecommons.org/images/stories/SAGEcomponents.jpghttp://www.sagecommons.org/images/stories/SAGEcomponents.jpg