RDFC2012 Open Access to Research Data

Open access to scientific research data
Gudmundur A. Thorisson, PhD <gt50@leicester.ac.uk>
Research associate, University of Leicester
Guest scientist, University of Iceland
Participant in the GEN2PHEN Consortium and the ORCID Technical Working Group

This work is published under the Creative Commons Attribution license (CC BY:
http://creativecommons.org/licenses/by/3.0/) which means that it can be freely
copied, redistributed and adapted, as long as proper attribution is given.

Overview

๏ Intro to the world of Big Science & Big Data
•Why is inadequate access to data such a problem?
๏ Incentive-based approaches to tackling the sharing problem
Identification, identification, identification
๏ Key relevant developments internationally
๏ Some food for thought for funders, institutions, other key players
๏ Concluding remarks

RDFC2012 Conference on Open Access and Digital Rights, Reykjavik, March 29 th 2012

Big Science, Big Data

• Scientific research increasingly large-scale and data-driven

• High-profile discipline examples

– High-energy particle physics - experiments
performed in the Large Hadron Collider

– Astronomy - data from ground-based and space
telescopes, the Virtual Observatory (VO)

• Doctorow, C. Big data: Welcome to the petacentre. Nature 455, 16-
21 (2008). http://dx.doi.org/10.1038/455016a

Hypothesis generation guided by available data

Kell and Oliver. Bioessays (2004) vol. 26 (1)

• Science paradigms
– 1st: Empirical - describing natural phenomena
– 2nd: Theoretical - models, generalizations
– 3rd: Computational - simulating complex phenomena
– 4th (1+2+3): Data exploration, e-Science

Gray, J. 2009. The Fourth Paradigm: Data-Intensive Scientific Discovery. Microsoft Research


Biological research too is
increasingly big and data-driven

• From: small-scale datasets that
fit into a printed journal article

Richards, M. et al. Paleolithic and neolithic lineages in the European mitochondrial gene pool. American
journal of human genetics 59, 185-203 (1996). http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1915109/


Biological research too is
increasingly big and data-driven

• To: large-scale collection of
biological data in digital form

• Huge technological advances in last 5-10 years
– experimental / observations <-- gathering data with high-throughput equipment
– computer technology <-- storing & analyzing massive data volumes

• Example: massively-parallel sequencing
– Determine human genome sequence in <1 day - the $1000 genome
– Metagenomics: sequence *everything* in environment samples
– Large bio-specimen collections
• x100,0000 of individuals in disease/population biobanks


Examples: domain repositories for sequence data

• GenBank - genetic sequence
repository, established 1986

• UniProt - knowledge base for
protein sequence & function
Conference on Conference on Open Access and Digital Rights, Reykjavik, March 29 th 2012
RDFC2012 Unique Identifiers, Vilnius, Feb 14 2012

“Community resource projects” - large-scale data generation
for the purpose of making the data available for broad reuse

• The sequence of the human genome
– International Human Genome project - mandatory rapid data sharing, the Bermuda
principles

• Pattern of variation in the human genome
– International Haplotype Map Project - genotyping population samples
– 1000 Genomes Project - sequencing population samples


Big Data – challenges, opportunities
• Managing & making sense of large-scale datasets
– Data easy/cheap to generate - not so cheap to store & use
– Favorite quote: “the $1000 genome sequence, followed by the ++$10,000 analysis”

• Integration & analysis - combining datasets
– more data of the same type - e.g. combine sequences from multiple species
– related data of different type - e.g. a person’s genome sequence + his/her phenotype

• Potential for accelerating research, creating new knowledge and (in
biomedicine) improving human health.

• Key driver = unrestricted sharing of scientifc data deposited in
the public domain


Data = “fuel” of science

Smith,V. Data publication: towards a database of everything. BMC Res Notes (2009) vol. 2 (1)


Data = “fuel” of science

[..] If digital technologies are the engine of this
revolution, digital data are its fuel. But for many
scientific disciplines, this fuel is in short supply.[..]

Smith,V. Data publication: towards a database of everything. BMC Res Notes (2009) vol. 2 (1)


Biology and data sharing in the “long tail”

• Biology is complex, so data are often very
heterogeneous
• Technologies changing rapidly
• Lots of small-scale research projects
• Lots of small/medium datasets The ‘long tail’ of dark bio-data
• Data in the long tail usually *not* shared
OR not shared in a useful way

• Contrast with other data-intensive disciplines with
– a long history of sharing research data - a “culture of sharing”
– big, expensive, shared facilities = the only way to do this kind of research
– relatively homogeneous datasets, easier to scale up to big volumes (e.g. telescope images)


[…] Overall, only 47 papers (9%) deposited full primary raw data
online. None of the 149 papers not subject to data availability
policies made their full primary data publicly available.

Conclusion: A substantial proportion of original research papers published in
high-impact journals are either not subject to any data availability
policies, or do not adhere to the data availability instructions in their
respective journals. This empiric evaluation highlights opportunities for
improvement


DATA
analysed
synthesised
interpreted

INFORMATION

published

KNOWLEDGE
Publication

Lots of published knowledge but
hard/impossible to go back and
reproduce work & validate findings

+
Opportunity for maximising the value of
data through reuse is wasted


Credit: http://cutcaster.com/photo/800902839-The-hand-drawing-question-WHY/

17

Lots and lots of diverse reasons!!
Some quotes from researchers:

“Don't which digital repository I should upload to”
“Too much work, got better things to do!”
“My competitors will just take the data and ‘scoop’ me”
“It's my data, I collected them and noone else is entitled
to use them”
“[myriad other reasons]”

Worringly, many authors don't seem to
care whether evidence underpinning their
published findings is accessible or not

Koslow. Should the neuroscience community make a
paradigm shift to sharing primary data?. Nat Neurosci
(2000) vol. 3 (9). http://dx.doi.org/10.1038/78760


Gnarly issue #1: “ownership” vs “stewardship”

• Many researchers consider data their property, even if research
funded by public money
– e.g. want to do further analysis on data in future, publish more papers

• ..which conficts with interests of other stakeholders in the game,
e.g. (funders, universities) who want:
– to maximize return on investment in the funded research
– to ensure good, solid evidence-based science is done, etc.


Gnarly issue #2 – biomedical data

• Usually sensitive, cannot be shared without restrictions
– Detailed, reidentifiable biomedical data that cannot be fully anonymized
– Personal privacy considerations

• Specialized controlled-access archives deal with some of this
– NCBI's database of Genotypes and Phenotypes – dbGaP
– European Genome-phenome Archive – EGA
– [specific diseases / disorders, research consortia, others]


How to Make a Tackle in Rugby
Tackling in rugby is one of the most important aspects of the game.
[...]
Credit:http://djamba.com/how-to-make-a-tackle-in-rugby.html

21

...which are an imperfect solution

• Arguments that mandates by themselves are not the way

• Mandates likely to ensure only minimum compliance
– sharing would be done in minimally useful form (as in, whatever is the least effort)

…. and are meaningless if not enforced (currently the case with
many journals)


Sharing now tends to be driven by mandates...

* Journals increasingly require data to be made available

“Provide supporting data in a repository OR we won’t
publish your paper”

* Funders increasingly require data sharing plan &
budget baked into grant proposals.

“Publish data we are funding you to generate OR we
will not fund your research again”

Using just a stick
gets you so only far


Strategies focused on encouraging sharing

- Make it easy -
- Make it useful -
- Make it citable -


Treating data as citable
publications in their own right

• Core strategy: enable data to be treated as 1st class citizens of the
scholarly record which:
i) are indexed and can be discovered, located and accessed, and
ii) can be properly identified & cited unambiguously like other scholarly works

• Link datasets with the primary journal publication - citation crosslinks
• Give data creators/curators/analysts proper credit for their contribution
to the digital resource

• Focus on the benefits to researchers from publishing their data
– Data sharing → Data PUBLICATION + CITATION
– Others reuse & cite their stuff → more citations → more impact
– The more useful a dataset, the more likely to be used & cited

Exemplar – Data Dryad

“international repository of data
underlying peer-reviewed articles in
the basic and applied biosciences”
http://datadryad.org

• Combines
– Mandates (journal policy)
and
– Citable data publication

• Citation cross-linking
– Paper references dataset
– Dataset references paper


Key building blocks: the 3 I’s of identification


1
I Identifying scholarly publications (and other research outputs)
• Why? So it is possible to..
..cite the work unambiguously (‘..we used the method described in Thorisson et al (2009)’)
..locate the work (retrieve Nature article as PDF from journal website)
..give credit to persons/entities who contributed to the work (G. Thorisson authored paper X)
• Need for globally unique, persistent identifiers to combat unstable Web URLs, broken hyperlinks
• e.g. Digital Object Identifiers (DOIs) for pubs, datasets and more:
– Bell et al. 2009. Science 323(5919) doi:10.1371/journal.pone.0024357
– Goodwillie C et al (2005) Data from: The evolutionary enigma of mixed mating systems in
plants: occurrence, theoretical explanations, and empirical evidence. Dryad Digital
Repository. doi:10.5061/dryad.292q34fp


2
I Identifying use/reuse - measuring impact

– Historical reliance on formal citations and citation-based metrics
– ISI Impact Factor widely used, but really metric for infuence of a scholarly journal
– Citation analysis not going away - remains the gold standard

– Many other use/reuse indicators for impact of individual research outputs
• Focus on the impact of the *publication* itself, not the journal in which it appears
• Indicators: no. full-text downloads, tweets (i.e. mentions on Twitter), social bookmarking
• AltMetrics - a growing grassroots movement “ to better measure and reward all the different
ways that people contribute to the messy and complex process of scientific progress [..] born out
of a simple recognition: Many of the traditional measurements are too slow or simplistic to
keep pace with today’s Internet-age science” http://altmetrics.org
– Lots new tools and projects emerging to explore possibilities in this space
• e.g. http://total-impact.org


3
I Identifying contributors – attributing credit

– Why? So we can..
..link content creators with their works - attribute credit accurately
..figure out: who contributed to publication X?
which publications has person/organization Y contributed to?
– What kind of contributions? Characterizing ‘contributorship’
author, creator, analyst, reviewer, ‘conceived of study & designed experiment’ etc


Tackling the author name ambiguity problem
(or ‘Who’s Who?’)

How about these? Or these?

J. Smith
J. Smith
J. Smith
Are these authors all the same person? J. Smith
G. Thorisson, University of Leicester J. Smith
G. A. Thorisson, University of Leicester [etc.]
G. A. Thorisson, Cold Spring Harbor Laboratory

∼2/3 of the ∼6 million authors in MEDLINE share a last name and
first initial with at least one other author, and an ambiguous name
refers to ∼8 persons on average.
Torvik and Smalheiser. Author name disambiguation in MEDLINE. ACM Transactions on Knowledge
Discovery from Data (2009) vol. 3 (3)


The Open Researcher & Contributor ID initiative

Launched end of 2009, ORCID will work to
support the creation of a permanent, clear
and unambiguous record of scholarly
communication by enabling reliable
attribution of authors and contributors
through unique identifiers


The Open Researcher & Contributor ID initiative

ORCID will add value for scholars and
the organizations that they are
interacting with, including universities,
scholarly societies, funding
organizations and publishers

•Joins faculty or student body
•Joins scholarly society
•Applies for grant
•Submits manuscript


ORCID transcends discipline, geographic, national and
institutional boundaries - now >300 participants

http://www.orcid.org 34

Some food for thought / recommendation
kind of stuff to conclude
• Status of research data in Iceland is unclear → need research
– Build on & extend 2007 Rannís report “Gagnagrunnar á Íslandi um náttúru, umhverfi og orku”

Rannís, we´re looking at you!

• Funders to take lead
– Mandates (aka sticks) - require data management plan + budget in grant proposals
• Many best practices & tools available to draw upon, e.g. by the UK Digital Curation Centre
– Call for & fund research proposals to build infrastructural foundations & explore
technologies/initiatives
– Raise awareness in the local research community


Even more food for thought /
recommendation kind of stuff

• Universities & other research institutions need to
– Take research data seriously
– Build infrastructure for data storage & preservation, support personnel (e.g. data
officers / coordinators)
– Include datasets and other non-conventional outputs in professional evalutations

• Identify & engage with key international initiatives in this space
– ORCID, DataCite, Dryad, Open Knowledge Foundation, others
– OpenAIRPlus ← Solveig's talk coming up!


Final bite of food-for-thought

Let's make research data an integral part of the
OA mission in Iceland, NOT an afterthought


Acknowledgements
GEN2PHEN Consortium
This work has received funding by the
http://www.gen2phen.org/about-gen2phen/partners European Community's Seventh Framework
Programme (FP7/2007-2013)
under grant agreement number 200754 -
Prof Anthony J. Brookes Bioinformatics Group, Leicester the GEN2PHEN project.

Contact me!
Contact me!
ORCID - http://www.orcid.org
<gthorisson@gmail.com>
<gthorisson@gmail.com>
http://www.linkedin.com/in/mummi
http://www.linkedin.com/in/mummi
http://www.twitter.com/gthorisson
http://www.twitter.com/gthorisson
Published under the Creative Commons BY license
http://www.gthorisson.name
http://www.gthorisson.name (http://creativecommons.org/licenses/by/3.0/)

RDFC2012 Open Access to Research Data

More Related Content

What's hot (18)

Viewers also liked (9)

Similar to RDFC2012 Open Access to Research Data (20)

More from Gudmundur Thorisson (15)

Recently uploaded (20)

RDFC2012 Open Access to Research Data