e-HTPX: A User Perspective
Robert Esnouf, University of Oxford
The e-HTPX Research Challenge
• Managing the processes along the protein
crystallography pipeline:
– Remote user managing a complete experiment
– Target selection, protein production, crystallization,
data collection, structure solution, deposition
– Exchange of data between laboratories,
synchrotrons, computational resources, etc…
– Development sites: SRS, EBI, York, Oxford, BM14
– Test sites: Oxford, York, Glasgow, (St Andrews’)
Potential benefits of e-HTPX
• Growing problems in protein crystallography:
– Higher throughput, automation, distributed working,
data volumes, better structure solution methods
• ESRF JSBG beam lines 74MB images, 1s exposures
– Need secure, universal data exchange
• Standardized data model, single sign on, universal naming
– Compute power
• Automated data analysis against structure/sequence DBs
– Current PX interest mainly limited to compute
resources and small clusters (MrBUMP)
e-HTPX e-Research Requirements
• e-HTPX is primarily a (meta-)data management &
data exchange problem:
– Administration: single sign on, access rights, roles
– Computation: small clusters (real-time data reduction)
– Data sharing & integration: standardization & unique
naming
– Workflow: ‘expert’ data collection software
– Collaboration tools: reporting, project management
Getting Users to Adopt e-HTPX
• Structural biology community is not necessarily
computer literate
– Focus more on understanding complex biology
– No local software installation / complex certificates
– Portal must offer real benefit in terms of experiment
automation, access to remote services or simplifying
data management/archiving
– Error prevention by exchanging metadata and using
barcoding intelligently
– Integrated access to tools and resources
Lessons from e-HTPX at Oxford
• Strengths
– Managing multi-researcher experiments
– Highly automated crystallization management
– ‘Champion’ driving developments at BM14
• Weaknesses
– Balance of flexible data model & simple UI
– Usefulness constrained by level of automation
– Added burden/restriction on experimental freedom
– Coping with rates of diffraction data acquisition
BM14 at the ESRF, Grenoble
Future Plans for e-HTPX
• Merging e-HTPX with other developments
– Widespread adoption of data exchange standards
– Crystallization management interface  PiMS
– e-HTPX Portal/Hub linked to PiMS (merged?)
– ISPyB now ESRF-wide standard experiment log
– DNA needs improving but desperately needed
– MrBUMP now part of CCP4 suite
– Diamond natural focus for synchrotron automation
Take home messages
• Crystallography is an experimental science
– The e is much less important than the Science
and must be invisible to the user
– A simple tool addressing a well defined problem is
much more likely to be taken up
– Uptake guaranteed if Diamond data management
is addressed
– Need science champions to get over adoption
‘transition state’
– Low data volume but high complexity & variability