2. Materials
Scientific research is +
generally held to be of
good provenance when Methods
it is documented in +
detail sufficient to allow
reproducibility. Results
=
http://en.wikipedia.org/wiki/Provenance#Science
Publication
3. In silico Experimental Standard Operating,
Procedures, Protocols, Plans
http://usefulchem.wikispaces.com/page/code/EXPLAN001
4. E. Science laboris: in silico experiments
Automated, reusable scripted
analysis pipelines - workflows
Data processing , data chaining
Data and tool integration
Annotation pipelines
Analytics
Simulation steering and parameter sweeps
Publication mining
Model and hypothesis building
Result Validation and comparison
Data cleaning, curation and preservation
Shield from clouds and
clusters details.
http://www.mygrid.org.uk/tools/taverna/
Record provenance: steps,
methods, results
5. Genetic variation in cattle species.
Food security, biodiversity.
Resistance to African trypanosomiasis infection
(sleeping sickness)
Liverpool (Kemp), Manchester (Brass), Nairobi
Comparing new data with reference genomes, prior results
and the literature to identifying interesting differences
22 million SNPs
6. Little Science +
Big Science
http://www.genomics.liv.ac.uk/tryps/trypsindex.html
7. Bottom up Effectiveness in Research
• Automated, repeatable, tracked plumbing
– Using institutional and community computing
infrastructures, tools and datasets
• Easier access to best of breed and “surfing” results
– non-developers access to sophisticated codes and
applications, shielded from nasty computing details.
• Leverage applications, services, datasets and codes
shielded from computing details.
– Honors original codes and applications. Heterogeneous
coding styles and tools sets. The best applications.
• Extensibility, adaptability & innovation.
– My stuff. Variant design.
8. Reuse, Recycle, Repurpose, Mash, Trade, Publish
Identify biological pathways implicated in
resistance to Trypanosomiasis in cattle using
mouse as a model organism.
Fisher P, et al Nucleic Acids Research, 2007, 35(16) 5625-5633
Dr Paul Fisher
Dr Jo Pennock
Identify the biological pathways involved in
sex dependence in the mouse
model, believed to be involved in the ability
of mice to expel the whipworm parasite.
Levison S.E., et al Inflammatory Bowel Diseases (2010)
10. How do I find and share
methods across the
institution, communities,
the web ?
How do I connect with
other authors and users?
How do I know if its any
good or right for me?
Who else is using it?
Where do I comment on
my experiences?
11. http://www.myexperiment.org
• Socially share, discover, review and reuse
workflows and other scientific methods.
• Cooperative market place.
• A scientific gateway.
• Commons-based Production + Social
networking
• Primary contribution, reviewing and curating.
12. Find experts and
peers, advice, work
flows, packs
Contribute, review
and curate
workflows
Train and educate
Launch workflows Cloud Methods Commons
http://www.myexperiment.org
13. Facts and Figures: Boutique but Beautiful
• Public Service: 1325 workflows, 349 files, 138 packs, 4129 registered
members, 235 groups, 56 different countries, ~ 3000 unique hits per
month. Workflows viewed/downloaded many 1000s of times.
• Adopted by 19 workflow systems and integrated into workflow
workbenches: Galaxy, Taverna
– Biology, chemistry, image analysis, social science, astronomy, engineering,
music…
– Specialised clones in Music & NeuroScience.
– Focus of research on workflow patterns and analytics.
• JISC funding since 2007
• (Other funding: Microsoft, EU, EPSRC, BBSRC)
14. Effectiveness and Open Collaboration
Open platform, off the shelf components, open
development, open linked data, Web 3.0 funky
Google gadgets
Application plugins
Linked Data Cloud
15. Workbench
Gadgets
Publishing
LogBook
Images
Software
Presentations
Literature
Compute resources
Backup and Archive Friends, colleagues, resources
[Duncan Hull] Data (files, spreadsheets)
17. Institutional
challenges
Adoption of Reproducible Methods
New Publishing and Learning Objects
Pre-and Post Publication Metadata Differentials
Citation, Credit and Reputation
Curation Costs
18. Methods Matter Science 2010
Reproducible
(or at least
defendable)
Research
many eyes
19. Virtual Learning
Actionable scholarly Environment
publishing & Undergraduate
Students
learning
Digital
Libraries scientists
Graduate
Students
Reprints
Peer-
Reviewed Technical
experimentation
Journal &
Conference Preprints Reports
&
Papers
Metadata
Local
Web Data, Metadata
Repositories Provenance
Certified
Experimental Workflows
Results & Analyses Ontologies
[De Roure]
23. Sharing Governance ….
“Its not ready yet”
“I need to get (another) publication first”
“We don’t have the resources or skills to prepare
it for others, esp. now we finished that project”
“Others won’t use it properly.” “Its not worth
my while”
“Its faster/easier to do it myself, and will get
the credit/control too”
“Its not described enough to be usable”
“I don’t trust the quality. Its not reliable enough. Its
too noisy.
25. Crowd
Contribution
Credit
Reward
Career
Use Profiles
Citation
Method
Building too.
Nature 2008
26. Credit and Reputation Community building
T Shirts are not enough
Coordination
Governance
QUALITY for REUSE
Sustainability
Software
Social & Cultural
Automation
27. Public Service
Open Software
Community generated
and curated Content
Dig Library/Repository
Social network
Collaboration platform
Computer science research
Software engineering
Teaching aid
Computational researchers Methodology
Researchers
Social Science Social experiment
29. Take Home: Methods Matter.
• Workflows are a transformative mechanism of
connecting tools and encoding know-how
– Scientists stand on the shoulders of resource experts
• myExperiment is a example of a collaborative
environment for connecting workflow authors and users
– Authors stand on the shoulders of each other
• The Power of Collectivism.
• Rewards and risks of researchers in competitive research.
• Cultural shift in reward, adoption and support for
building, sharing and curating computational methods.
30. Acknowledgements
myExperiment Director: David De Roure
Developers Users Sponsors
• Jiten Bhagat • Katy Wolstencroft • Savas Parastatidis
• Don Cruickshank • Paul Fisher • Roger Barga
• Danius Michaelides • Duncan Hull • Derick Campbell
• David Newman • Franck Tanoh • Tony Hey
• Sergejs Aleksejevs • Andrea Wiggins
• Mark Borkum • Marco Roos
• Matt Lee • Jerzy Orlowski
• Tom Foster • Olga Krebs
• Wolfgang Mueller
Allied, Contributing Projects • Tony Linde
• Thomas Laurent
• Eric Nzuobontane Social Scientists
• Ian Dunlop • Yuiwei Lin
• Stuart Owen • Rob Proctor
• Shoaib Sufi • Meik Poschen
• Sean Bechhofer • Jonathan Foster
• Rodrigo Lopez
• Steve Pettifer
• Mannie Tags
• Finn Bacall
• Sarah Thew
• Matt Gamble
• Tim Clark
31. Contact
David De Roure
dder@ecs.soton.ac.uk
Carole Goble
carole.goble@manchester.ac.uk
Visit myexperiment.org