benefits of open science lennart martens lennart.martens@vib-ugent.be computational omics and systems biology group VIB / Ghent University, Ghent, Belgium CC BY-SA 4.0 Open science creates very many opportunities but it does take some work to get it right • Open science should not serve to police scientists, but to provide opportunities • Creating an open data ecosystem is not trivial, but with a bit of work, it can certainly be done • Scientists should not ask themselves whether there will be open science, but rather what they will be able to do thanks to open science CC BY-SA 4.0 Why should we share our work? Our biggest challenge is to make it work! What can we do with open science? Of sedimentation, opportunity, and (an absence of) dragons CC BY-SA 4.0 Why should we share our work? Our biggest challenge is to make it work! What can we do with open science? Of sedimentation, opportunity, and (an absence of) dragons CC BY-SA 4.0 We usually think we need open science to prevent bad things from happening • While open science helps prevent some cases of fraud or low quality work being published, it is certainly not a panacea (cfr. peer review) • Simultaneously, fraud is regularly detected: • in the absence of the source data • from papers published in closed access journals • without any of the code or metadata available • Why should we define the use of open science through an application with negative connotation? CC BY-SA 4.0 Instead, we should rather focus on the good that comes from open science • Open science makes the work accessible to anyone • Open science allows people to build much more efficiently on previous work • Open science helps maximize the usefulness of each individual research effort • Data tend to have a (much!) longer shelf life than our (limited) interpretations • Open science fosters creativity, and stimulates revolutionary research CC BY-SA 4.0 Why should we really have open science? Our biggest challenge is to make it work! What can we do with open science? Of sedimentation, opportunity, and an absence of dragons CC BY-SA 4.0 Making open science work requires a bit of effort from every scientist • The data that is obtained should be accompanied by the associated metadata • The code that is written should be understandable, documented, and hosted at a reliable site • The protocols should be provided clearly and in full • The interpretations should be clearly linked to the data (full provenance) • Everything should be licensed in a permissible way CC BY-SA 4.0 Any open data exchange ecosystem requires standardization Masuzzo, Trends in Cell Biology, 2014 CC BY-SA 4.0 But scientists are human, and very fallible - especially when extra effort is required Verheggen, EuPA Open Proteomics, 2015 CC BY-SA 4.0 But scientists are human, and very fallible - especially when extra effort is required Verheggen, EuPA Open Proteomics, 2015 CC BY-SA 4.0 The arrival of user-friendly submission tools did not nothing to reverse reporting trends Verheggen, EuPA Open Proteomics, 2015 CC BY-SA 4.0 Manual curation of submissions, equivalent to restrictive policing, did help Verheggen, EuPA Open Proteomics, 2015 CC BY-SA 4.0 Mid-to long-term storage of data is expensive and thus very difficult to fund and maintain http://www.ncbi.nlm.nih.gov/peptidome Slotta, Nature Biotechnology, 2009; Csordas, Proteomics, 2013; Martens, Proteomics, 2013 CC BY-SA 4.0 And as responsible caretaker of your stuff, you will sometimes need to take action too http://peptide-shaker.googlecode.com Vaudel, Nature Biotechnology, 2015 CC BY-SA 4.0 Maintaining what you do is not trivial, not cheap, and considered unwise by senior PIs http://peptide-shaker.googlecode.com Vaudel, Nature Biotechnology, 2015 CC BY-SA 4.0 Why should we really have open science? Our biggest challenge is to make it work! What can we do with open science? Of sedimentation, opportunity, and an absence of dragons CC BY-SA 4.0 The PRIDE database was started to allow (orthogonal) re-analysis of proteomics data CC BY-SA 4.0 Martens, EuPA Open Proteomics, in press A large amount of post-consumer MS data is collected in public databases such as PRIDE CC BY-SA 4.0 This is where I left Martens, Proteomics, 2005 The identified peptides reported by the community proved highly informative Foster, Proteomics, 2011; Colaert, Nature Methods, 2011; Barsnes, Proteomics 2011, Vandermarliere, Proteomics 2013; Degroeve, Bioinformatics 2013 CC BY-SA 4.0 Meanwhile, we took an interest in the unidentified part of the data CC BY-SA 4.0 We built the ReSpin pipeline to enable fast re-processing of proteomics data in new ways Experiment PRIDE Re-analysis Original identifications ReSpin New knowledge CC BY-SA 4.0 First of all, we had to understand the data to allow reliable re-interpretation http://compomics.github.io/projects/pride-asa-pipeline.html Hulstaert, Journal of Proteomics, 2013 CC BY-SA 4.0 Then we had to make it easy to re-analyze these data with multiple search algorithms http://compomics.github.io/projects/searchgui.html Vaudel, Proteomics, 2011 CC BY-SA 4.0 And finally, a means to collate, process and validate the results from our re-analyses http://compomics.github.io/projects/peptide-shaker.html Vaudel, Nature Biotechnology, 2015 CC BY-SA 4.0 Incidentally, anyone can reprocess public data with PeptideShaker and about ten mouseclicks CC BY-SA 4.0 Vaudel, Nature Biotechnology, 2015 We then automated everything on our Pladipus custom grid engine http://compomics.github.io/projects/pladipus.html Verheggen, Journal of Proteome Research, 2016 CC BY-SA 4.0 Combining data sets brings clear benefits Vizcaíno, Nature Biotechnology, 2014; Wilhelm, Nature, 2014; Kim, Nature, 2014 (Kuester + Pandey) – PRIDE CC BY-SA 4.0 We already went hunting for translated lncRNAs, but we found very few (< 1%) Volders, NAR, 2013; Volders, NAR, 2015 lncRNA Forward Decoy CC BY-SA 4.0 We were able to help confirm expression of small ORFs across human tissues Olexiouk, NAR, 2016 CC BY-SA 4.0 Data growth is increasing in PRIDE; more, and ever bigger data sets are submitted daily Number of submissions permonth Vaudel, Proteomics, 2015 CC BY-SA 4.0 Projections for data growth in PRIDE are therefore quite impressive Courtesy of Dr. Juan Antonio Vizcaíno, Proteomics Team Leader, EMBL-EBI CC BY-SA 4.0 All of the available data simply bristle with opportunity Vaudel, Proteomics, 2015 CC BY-SA 4.0 Why should we really have open science? Our biggest challenge is to make it work! What can we do with open science? Of sedimentation, opportunity, and an absence of dragons CC BY-SA 4.0 A sociologist’s take on our efforts towards (orthogonal) data reuse “This desire to reactivate data is widespread, and Klie et al. are not alone in wanting to show that ‘far from being places where data goes to die’ (Klie et al., 2007: 190), such data collections can be mined for valuable information that could not be obtained in any other way.” “In attempting to reactivate sedimented data in order to enable its re-use, their first step was ...” "... they are experiments in seeing, in furnishing ways of seeing how data on proteins could become re-usable, could be reactivated as collective property rather than the by-product of publication." Mackenzie and McNally, Theory, Culture and Society, 2013 CC BY-SA 4.0 One of the big ideas in science will be the (ortogonal-) re-use of (big) public data Data repository And now think about open science, and imagine the opportunities • What could you do with open science? What could you study? What could you learn? • What opportunities would present themselves, if… • All data (in your field) were available online • All algorithms (in your field) were available online • All publications (in your field) were open access • Most of these opportunities are not little steps forward; instead they promise to be revolutionary! CC BY-SA 4.0 J.R.R. Tolkien, A Conversation with Smaug https://www.flickr.com/photos/fantasy-art-and-portraits/2884954207 (CC BY-NC-SA 2.0) Here is treasure of unlimited size, with all dragons chased away – now what will you do? CC BY-SA 4.0 CC BY-SA 4.0 www.compomics.com @compomics CC BY-SA 4.0