The NCIP is committed to advancing cancer research through improving biomedical informatics, and we believe that active participation by the broad community is the most effective way to accomplish this. To that end, we have made significant advances in our plan to engage the biomedical-research, informatics, and developer communities by lowering the barrier to community-driven software development.
In a previous blog post about NCIP’s plans to move into open-development published on July 26, 2012, Juli Klemm reflected on different ways for us to “get there from here.” Since that time, we have made tremendous progress in driving our strategy forward and are excited to announce that we have tackled the main recommendation that came out of our June meeting with open-source thought leaders: Keep it simple. Make barriers to entry as low as possible, and reuse available resources. Specifically, we have adopted a software license that is approved by the Open Source Initiative (OSI) and have begun to migrate the code developed under the cancer Biomedical Informatics Grid® (caBIG®) Program to a public repository. Our goal in taking these steps is to remove as many barriers as possible to community participation in the continuing development of these assets.
At the time the caBIG® license was created, the goal was to ensure that no restrictions would be placed upon anyone who wanted to make use of caBIG® code. That license was permissive, which means that derivative works would not be required to be under a compatible license. This lack of restriction is a significant enticement to commercial vendors, who may wish to protect their derivative work. Our goal was then, and still remains, to encourage both open-source and commercially viable derivatives of any work created under the caBIG® Program and, going forward, the NCIP. Another important aspect of the caBIG® license was that it did not allow endorsement of derivative works using the names of the contributors. As an entity of the U.S. government, both aspects of the original license remain important to us.
Given these considerations, we gravitated to the BSD 3-Clause License. This is a very simple, well-understood, OSI-approved, permissive license that does not allow endorsement. The simple form of this license should not create any impediment to anyone who wishes to make derivative works from the code. Thus, the BSD 3-Clause License meets all of our key requirements. We are currently in the process of inserting this license into all of our existing code files and will use it with all future NCIP work products.
While caBIG® code has always been available for anyone to download from our website, we did not have a simple mechanism for third-party code contributions to be merged into our code base. To resolve this issue, we have already begun to migrate the bulk of this software code to an NCIP channel in GitHub. GitHub is a widely used code-hosting site that offers collaboration tools, as well as document-management, issue-tracking, and code-management capabilities, all features that should empower the community to participate in maintaining and improving these resources.
Our hope is that the availability of this simple, open license and the migration of our code to GitHub will encourage both end-users and developers to coalesce around projects, develop vibrant communities of interest and practice, and create a positive overall user experience for everyone. As a result of this new community-led development approach, we will discover which projects meet the greatest needs of the community by observing their level of uptake. Those projects around which vibrant communities form will be most likely to attract the attention of NCI funding programs. More importantly, a vibrant community will ensure more rapid advances in biomedical informatics in support of cancer research.
To date, seven applications have been migrated to GitHub: the Patient Study Calendar (PSC); caArray; caIntegrator; cancer Bench to Bedside (caB2B); cancer Genome-wide Association Studies (caGWAS); caBIG® Central Clinical Participant Registry (c3PR); and cancer Laboratory Information Management System (caLIMS). More migrations are being queued up now, and we will keep you posted on the NCIP Twitter feed, the NCIP LinkedIn group, and the NCIP Announce listserv as soon as those applications have been moved. Please don’t hesitate to let us know your thoughts; we are eager to hear feedback on our efforts.
Anthony R. Kerlavage, Ph.D., is Chief, Scientific Programs at the Center for Biomedical Informatics and Information Technology (CBIIT), National Cancer Institute (NCI). You may connect with Tony via LinkedIn.