GigaScience EiC Laurie Goodman's talk at the Council of Science Editors Annual Meeting on "Laurie Goodman at #CSE2014: Reproducibility: It's going to cost you time and effort, but it's our job!". 4th MAy 2014
2. Growing Issue: increasing number of retractions
>15X increase in last decade
Strong correlation of “retraction index” with
higher impact factor
1. Science publishing: The trouble with retractions http://www.nature.com/news/2011/111005/full/478026a.html
2. Retracted Science and the Retraction Index ▿ http://iai.asm.org/content/79/10/3855.abstract?
At current % increase by 2045 as
many papers published as
retracted!
3. Problem: Growing Replication Gap
1. Ioannidis et al., (2009). Repeatability of published microarray gene expression analyses. Nature Genetics 41: 14
2. Ioannidis JPA (2005) Why Most Published Research Findings Are False. PLoS Med 2(8)
Out of 18 microarray papers, results
from 10 could not be reproduced
4. The problems with publishing
• Scholarly articles are merely advertisement of scholarship .
The actual scholarly artefacts, i.e. the data and
computational methods, which support the scholarship,
remain largely inaccessible --- Jon B. Buckheit and David L.
Donoho, WaveLab and reproducible research, 1995
• Core scientific statements or assertions are intertwined and
hidden in the conventional scholarly narratives
• Lack of transparency, lack of credit for anything other than
“regular” dead tree publication
5. GigaSolution: deconstructing the paper
Need to credit and reward:
• Data/software availability
• Metadata/curation
• Interoperability
• Availability of workflows
• Transparent analyses
Data
Metadata
Methods
Analyses
Every Component
Has a Citable DOI
6. How are we supporting reproducibility?
Data Sets in
GigaDB
Analyses in
GigaGalaxy
Paper in
GigaScience
Open-access journal Data Publishing Platform
Data Analysis Platform
7. Example in Neuroscience
• Neuroscience Data are
not typically shared
• Data AND Tools are not
typically made
available to the
reviewers
• Journal Editors think
Reviewers will not want
to review data
GigaScience 2014, 3:3 doi:10.1186/2047-217X-3-3
8. Example in Neuroscience
• Neuroscience Data are not typically shared
• Author Dr. Stephen Eglen said: “One way of encouraging neuroscientists to
share their data is to provide some form of academic credit.”
• We hosted with a DOI: 366 recordings from 12 electrophysiology datasets
• GigaDB is included in Thompson Reuters Data Citation Index
• Data AND Tools are not typically made available to the reviewers
• We made manuscript, data and tools all available to the reviewers.
• We make sure to include reviewers who are able to properly assess the data
itself and rerun the tools
• To reduce burdens- we sometimes select a reviewer who ONLY looks at the
data.
• Journal Editors think Reviewers will not want to review data
• What Reviewer Dr. Thomas Wachtler said: “The paper by Eglen and
colleagues is a shining example of openness in that it enables replicating the
results almost as easily as by pressing a button.”
• What Reviewer Dr. Christophe Pouzat said: “In addition to making the
presented research trustworthy, the reproducible research paradigm
definitely makes the reviewers job more fun!”
9. Reviewer Comments on the Process
http://www.biomedcentral.com/biome/christophe-pouzat-and-thomas-wachtler-on-reproducible-research-in-neuroscience/
http://blogs.biomedcentral.com/gigablog/2014/04/16/qa-on-dynamic-documents/
10. Reviewer Take Home Message
http://www.biomedcentral.com/biome/christophe-pouzat-and-thomas-wachtler-on-reproducible-research-in-neuroscience/
http://blogs.biomedcentral.com/gigablog/2014/04/16/qa-on-dynamic-documents/
11. But This Can’t Be
Done For Wet Bench!!!
And Make All Data Available
Take steps to put all the images and raw data in a repository,
and to use biobanking where possible
Address Insufficient-Method
Syndrome:
-Complete, Easily found,
Searchable, and Updatable
12. Lessons learned:
• It is possible to recreate a result from a paper
• Reproducibility is COSTLY. How much are you
willing to spend?
• Learn a huge amount about the study, and provides lots
of information not present in the paper
• Much easier to do this before rather than after
publication
13. It’s Time to Move Beyond
Dead Trees
18121665 1869