When I started my PhD I was rather naïve about the workings of academia, never mind open science. This is my story on how I learned that open science, and in my case open source publishing, can not only fast-track science itself but actively stimulate collaboration, maximize impact, and in the process boost your career.
The topic of my PhD (2006-2012 TCD) was the creation of an experimental and computational framework to study the biomechanical properties of human soft tissue (relevant to the improvement of car safety through simulations of tissue injury during crashes). The basic set of tools I required involved visualization, image processing, 3D modeling, and computational analysis. For each it seemed separate commercial and closed source software was available requiring specific licenses and license servers. Some software was simply not within budget, while others only operated on specific sites on campus and only if enough shared licenses happened to be available. In addition, to perform all the steps I required, I had to manually export/import files between packages. The inefficiency of this process and the costs associated with the commercial licenses frustrated me. It was also frustrating that I could not see how these systems worked under the hood. I wanted to understand and validate the software components and expand their capabilities, automate them, and integrate them into a single efficient pipeline. I therefore set out to create a dedicated toolbox within the programming environment MATLAB capable of performing all steps in a clear and transparent way (see visual examples in Figure 1).
When I started my first post-doctoral position (2012-2015 at the AMC) it became clear that others were suffering from the same frustrations and I began to informally share codes with colleagues. Keeping track of versions quickly became an issue so I posted the toolbox on the web-based repository GitHub, which offers version management and is free for public projects. So, in my case, posting the codes online stemmed from a necessity to manage the project on a free and web-based platform. I also needed to decide on a type of license. Since I had little experience with publishing open source, I was at first wary of sharing the codes online. Naively, part of me felt that the work would have little impact, that people would somehow steal my work and that therefore I would not be credited for it. I decided to publish the codes with an open source license (to read more about different types of open source licenses see here and here), and quickly learned my worries could not have been further from the truth.
As a way to advertise the toolbox I presented it at the international computational methods conference CMBBE 2013 (Kevin M Moerman, Nederveen, and Simms 2013). To give the toolbox a more official feel, I created a website and named the toolbox GIBBON: The Geometry and Image-Based Bioengineering Add-ON. To further improve visibility I made the codes available on a popular MATLAB code sharing site causing the number of downloads to increase significantly. The following year at CMBBE 2014 I organized a free workshop on GIBBON as well as co-chairing a special session on computational modeling. Through this I was able to attract international users and GIBBON started to receive approximately 50 downloads a month. So by advertising the work I removed the little impact worry. Addressing the not being credited aspect was even easier since through the research sharing initiative Zenodo I assigned a DOI to the GitHub project allowing users to start citing my work (Kevin Mattheus Moerman 2014).
By publishing software as open source, researchers can immediately copy and start integrating and customizing the codes for their own use. I have personally enjoyed these aspects by integrating and linking GIBBON with two other powerful open source projects; TetGen, and FEBio. The ease of copying and/or editing should not be seen as threat (i.e. people stealing my work), since licenses can be put in place that ensure due acknowledgement and citation, but as a way to facilitate and accelerate the advances in the field of computational science. In addition, web-repositories, such as GitHub, allow users to create feature requests (keeping development on your side), or to feed altered codes back into your project making them potential contributors, thus actively stimulating collaboration.
The institute where I did my PhD maintains an interest in GIBBON, and my continued collaboration there has led to me being appointed as a visiting research fellow. I have since coauthored several publications with this group (Cooney et al. 2015; Takaza et al. 2012; Gindre et al. 2013).
Another collaboration was kick-started when I was approached by David Sengeh, a PhD student at the MIT Media Lab. David now uses GIBBON for prosthetic socket design optimization for which I added many new features. This fruitful collaboration let me to start co-supervising David. This collaboration led to a job offer and in November 2015 I will start a Post-Doctoral Associate position at the MIT Media Lab. Several journal publications are currently being prepared in relation to this work, and some preliminary results have already been presented by David Sengeh in his TED talks here and here.
My personal application for GIBBON has mainly been muscle mechanics ((Kevin M Moerman et al. 2011; Kevin M. Moerman et al. 2010; Kevin M. Moerman et al. 2013; Kevin M. Moerman et al. 2012)(Kevin M. Moerman et al. 2010)) however the open source access to GIBBON has meant it is now also finding application in pressure ulcer mechanics (Jagt 2014), image inversion theory (Moosbauer 2015), and soon the study of blood vessel, and bone mechanics.
Publishing GIBBON open source has maximized the impact of my work on an international level, and has resulted in the enhancement of my career through collaborations, citations, co-authored publications, and ultimately, a job at one of the most prestigious institutions for my field of research.
Figure 1 The GIBBON logo (centre) and visualizations for various applications
Cooney, Gerard M, Kevin M Moerman, Michael Takaza, Des C Winter, and Ciaran K Simms. 2015. “Uniaxial and Biaxial Mechanical Properties of Porcine Linea Alba.” Journal of the Mechanical Behavior of Biomedical Materials 41 (January): 68–82. doi:10.1016/j.jmbbm.2014.09.026.
Gindre, Juliette, Michael Takaza, Kevin M. Moerman, and Ciaran K. Simms. 2013. “A Structural Model of Passive Skeletal Muscle Shows Two Reinforcement Processes in Resisting Deformation.” Journal of the Mechanical Behavior of Biomedical Materials 22 (June). Elsevier: 84–94. doi:10.1016/j.jmbbm.2013.02.007.
Jagt, Thyrza. 2014. “Modeling the Interaction between Micro-Climate Factors and Moisture-Related Skin-Support Friction during Patient Repositioning in Bed.” Technical University of Delt. http://ta.twi.tudelft.nl/nw/users/vuik/numanal/jagt_scriptie.pdf.
Moerman, Kevin M, Aart J Nederveen, and Ciaran K Simms. 2013. “Image Based Model Construction , Boundary Condition Specification and Inverse Fea Control : A Basic Matlab Toolkit for Febio,” 7–8.
Moerman, Kevin M, Andre M J Sprengers, Ciaran K Simms, Rolf M Lamerichs, Jaap Stoker, and Aart J Nederveen. 2011. “Validation of SPAMM Tagged MRI Based Measurement of 3D Soft Tissue Deformation.” Medical Physics 38 (3): 1248–60. doi:10.1118/1.3533942.
Moerman, Kevin M., Christian M. Kerskens, Caitríona Lally, Vittoria Flamini, and Ciaran K. Simms. 2010. “Evaluation of a Validation Method for MR Imaging-Based Motion Tracking Using Image Simulation.” Eurasip Journal on Advances in Signal Processing 2010 (Article ID 942131): 11 pages. doi:10.1155/2010/942131.
Moerman, Kevin M., Andre M J Sprengers, Aart J. Nederveen, and Ciaran K. Simms. 2013. “A Novel MRI Compatible Soft Tissue Indentor and Fibre Bragg Grating Force Sensor.” Medical Engineering and Physics 35 (4). Stockholm, Sweden: 486–99. doi:10.1016/j.medengphy.2012.06.014.
Moerman, Kevin M., Andre M. J. Sprengers, Ciaran K. Simms, Rolf M. Lamerichs, Jaap Stoker, and Aart J. Nederveen. 2012. “Validation of Continuously Tagged MRI for the Measurement of Dynamic 3D Skeletal Muscle Tissue Deformation.” Medical Physics 39 (4): 1793. doi:10.1118/1.3685579.
Moerman, Kevin Mattheus. 2014. “GIBBON (Hylobates Agilis).” doi:10.5281/zenodo.12214.
Moosbauer, Christoph. 2015. Inverse Problems In Dynamic Elasticity Imaging. Anchor Academic Publishing (aap_verlag). https://books.google.com/books?hl=en&lr=&id=jjafBwAAQBAJ&pgis=1.
Takaza, Michael, Kevin M. Moerman, Juliette Gindre, Garry Lyons, and Ciaran K. Simms. 2012. “The Anisotropic Mechanical Behaviour of Passive Skeletal Muscle Tissue Subjected to Large Tensile Strain.” Journal of the Mechanical Behavior of Biomedical Materials 17: 209–20. doi:10.1016/j.jmbbm.2012.09.001.
Showing 2 Reviews
This is a nice overview of the journey that a researcher typically takes from a very basic prototyping, to high-quality massively reusable piece of software, mirroring the researcher's career path more generally. The citations and download data support the overall conclusions about the benefits of sharing the source code openly. I particularly like the part explaining how the availability of the code made it possible for the software to evolve independently and find its applications further afield from what the author has originally envisaged. I am sure Kevin will find how, with time, the applications are getting even more 'out of control' and may be surprised to see the code find its way in areas completely unrelated to bioengineering. It would be interesting to see a few words on further benefits from active promotion, e.g. via social media channels.
I really liked the evolution from feeling that open sharing would have "little impact", that people would "steal the work" and that there would "not be credit" for it, to the recognition that in fact, the open sharing created an environment that was exactly the opposite. Also important is the journey for a software developer to become open, to take advantage of some of the excellent tooling available to assist in open collaboration. Github really is changing how we develop open software. One thing I was wondering, is how you have managed tracking usage besides the Github ticket tracker/code contributions, as this is one area of impact/attribution that is often very difficult in open source projects.
4pts for "inspiration factor"
5pts for compelling nature of evidence
4pts for quality of writing
5pts for impact of activities on personal goals, science, or scholarship.
This article and its reviews are distributed under the terms of the Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and redistribution in any medium, provided that the original author and source are credited.