Overview

1
Overview

• Question
• Does collaborative technology enhance the
  productivity and effectiveness of biomedical
  research collaborations engaged in software
  development?
• Approach
• Select promising tools and evaluate their
  usefulness and impact in computational research
  testbeds. Iterate.
• Testbed areas
• Structural biology (UCSF, Scripps, PSC, NIEHS
  AMBER, CHARMM)
• Computational pathology (Pitt, UPMC, NCI)
• Visible Human (PSC, UMich expanding to Ferris
  State and Duke)
• NCRR Supplement (1998-2002)
• Biomedical Collaboratories Workshop October
  27-29, 2000

2
Does collaborative technology enhance the
productivity and effectiveness of biomedical
research collaborations?

• Collaboratory Results
• CVS distributed source control, electronic
  notebooks, Net/SGImeeting cross-platform
  application sharing, and chat tools improved the
  efficiency and quality of code design/implementati
  on/validation and the access/utilization of codes
• Web-based chat tools and Net/SGImeeting
  cross-platform application sharing supported need
  for immediate response within ad hoc
  communication patterns during distributed CHARMM
  runs under LEGION Grid Computing Infrastructures
• Incorporating collaborative features into the PSC
  Volume Browser tool enabled instructor-led and
  collaborative exploration of visible human data
  expanding to include segmentation and multiple
  volumetric datasets
• Technological Issues
• Tools require significant change in work process
• Usability User training and support is vital
• Teleconferencing tools are valued but challenging
  (e.g., heterogeneous platforms, ad hoc
  communication patterns, workday offsets)
• SysAdmin/Networking issues (ports?)
• Always one application missing. J. Myers
• Social/Cultural Issues
• Attributes of the collaboration itself affect the
  implementation and use of collaboratory tools,
  e.g., maturity, pattern of operations, structure,
  and funding
• Must add value to collaborating scientific
  community

3
Key Accomplishments

• CVS distributed source control, electronic
  notebooks, Net/SGIMeeting cross-platform
  application sharing, and chat tools improved the
  efficiency and quality of code design/implementati
  on/validation and the access/utilization of
  codes.
• Web-based chat tools and Net/SGIMeeting
  cross-platform application sharing supported the
  need for immediate response within ad hoc
  communication patterns during distributed
  simulations across the Legion Computational Grid,
  such as CHARMM protein folding energy landscape
  exploration in Structural Biology.
• Our focus on assessing the utility of existing
  and emerging generally available collaborative
  computing technologies, rather than
  custom-developed tools, in our diverse user
  communities revealed that many of these tools are
  not yet adequate to meet their needs, e.g.,
  Net/SGIMeeting problems with firewalls, other
  security issues, and inadequate application
  sharing of sophisticated 3D image manipulation.
  These deficiencies were often the motivation for
  custom-developed collaboratory tools or
  collaborative functionality within existing
  tools.
• Incorporating collaborative features into the PSC
  Volume Browser tool enabled instructor-led and
  collaborative exploration of visible human data.

4
Suggestions for future study...

• Grid technology is becoming an important
  collaboratory technology and is a suitable
  subject for further study. It would be worthwhile
  to develop and assess tools that facilitate
  practical implementation of Grid capabilities.
• Evaluation continues to be seen as an important
  motivator of future funding of collaborative
  efforts and there is always pressure for harder
  results, e.g., use of controls or gold
  standards.
• When the collaborative technologies are
  inextricably linked to the collaboration itself,
  additional methodological complexities hold.
  Studies that address these questions would make
  important contributions to our understanding of
  value-added
• To what are you comparing collaboratory
  experiences? (e.g., remote access using
  traditional methods, onsite use of the resource)
• What are appropriate measures of success if we
  dont know what the outcome would have been
  without the intervention of the tools?
• Is it easier to evaluate a collaboratory when the
  research task cannot be accomplished without
  collaboration? Or is it harder because the
  baseline and desired end-state are so different?