Computational Science Takes a Village - How Code Ocean Enables Internal Collaboration

Efficient collaboration is a key factor for success [1, 2]. While the term collaboration often pertains to working with outside partners, collaboration really starts “at home” with successfully sharing and working with members of the same or a cross-functional team.

In an ongoing series of blogs under the heading “Computational Science Takes a Village”, we are sharing how the Code Ocean Computational Workbench and Hub can facilitate collaboration between team members, across research teams within a company and with outside partners. In this blog we share an example of how Compute Capsules(™) can jump-start computational analysis and how an expert user drove the broad adoption of a Compute Capsules to accelerate time to results.

Getting Started More Quickly

Most computational scientists are not experienced software engineers and frequently need help not just with coding but with setting up analysis pipelines. Tasks like:

• selection of analysis packages and versions,
• building and managing a stable computing environment,
• setting up the necessary security protocols and access privileges,
• setting up repos, e.g. in GitHub,
• making sure the data are provided in the correct format, and
• adding app panels and visualization tools that enable broad use

require skills computational researchers are often not trained in. As a consequence they turn to colleagues (or IT) with more coding experience and proficiency which can lead to delays and slipping time lines, especially if the experienced software engineers are in high demand and become a bottleneck.

The problem is exacerbated when the original developer of the analysis pipeline wants to share their work with colleagues. To use the analysis tool every user needs to make the same code and coding environment work on their computing machines – again requiring the help of the limited experts.

As a result, not only can the start of a project be significantly delayed, but also collaboration between and among teams can be negatively impacted, valuable analysis tools cannot be fully leveraged across the organization and tools, datasets and whole analyses can be lost, e.g. when an employee leaves, because they cannot be used by others.

Compute Capsules – Fast Collaboration Made Easy

Code Ocean Compute Capsules were specifically designed to solve this exact challenge. Compute Capsules are software packages that allow researchers to add code, data, computing environment and results to create a self-contained research asset that can then be easily shared with others. These additional users simply access the Compute Capsule with their credentials and are immediately up and running. While the initial set-up might still require input from experienced coders, this is a one-time engagement; all additional users automatically work in the standardized Capsule environment with no need to customize any aspect of their computing environment.

Compute Capsules therefore reduce the demand for often scarce software engineering and compute resources and allow computational scientists to get started faster. In addition, Compute Capsules have a second critical advantage:

Collaboration with Compute Capsules Drive Productivity Across the Organization

A new analysis pipeline will be implemented, used, shared and built on more quickly and frequently if it is easy to adopt. Code Ocean Compute Capsules lower the barrier to broad adoption significantly because – except logging in – no additional steps are required by new users. Based on that easy access a Compute Capsule containing a highly useful computational pipeline will experience rapid adoption throughout an organization. Not only can the original results now be easily reproduced, but also new data sets can be quickly analyzed and derivative Capsules can be created building on and broadening the original use.

The following example from a Code Ocean customer illustrates how quickly a successful Capsule can spawn broad adoption across a project team within a company.

A power user, the company’s Lead Data Scientist, built an analysis tool for a bioinformatician who then needed the help of a senior data scientist. With his help, an algorithm developer and the head of research modified and used the Capsule to run similar analyses. These Capsules lead to a total of 18 project team collaborations over a seven-month period.

Code Ocean’s goal is to make computational science more open, accessible and collaborative through Compute Capsules. Compute Capsules make it easier and less resource intensive to get analysis work started and facilitates broad interdisciplinary adoption and collaboration.

To learn more and discuss how your company could benefit from Code Ocean’s Computational Workbench Please get in touch.

References

[1] Geraei, E., Mazaheri, E., & Karimi, M. (2018). Intradepartment scientific collaboration in Journal of Research in Medical Sciences: A co-authorship study. Journal of research in medical sciences : the official journal of Isfahan University of Medical Sciences, 23, 97. https://doi.org/10.4103/jrms.JRMS_36_18

[2] He, Z. , Geng, X., Campbell-Hunt, C. (2009). Research collaboration and research output: A longitudinal study of 65 biomedical scientists in a New Zealand university. Research Policy, 38, 2.
https://doi.org/10.1016/j.respol.2008.11.011.