The National Cancer Institute's National Cancer Informatics Program is adopting a new open source license model and migrating its code to the GitHub code-hosting site in hopes of making it easier for academic institutions and commercial firms to download and make changes to software and code developed by projects funded by its Center for Biomedical Informatics and Information.
Specifically, NCIP has adopted the BSD 3-Clause, which is approved by the Open Source Initiative, to replace an existing open source license that governs access to and use of the Cancer Biomedical Informatics Grid infrastructure.
NCIP inherited caBIG last April following the NCI's reassessment of the program based on recommendations made by its Board of Scientific Advisors. The NCI decided to scale back and eliminate some caBIG projects and handed responsibility for the surviving programs over to NCIP (BI 4/20/2012).
Anthony Kerlavage, the chief of CBIIT's scientific programs, told BioInform that the new license is essentially identical to the previous caBIG license, which caBIG drafted itself in 2005 (BI 4/18/2005), but the BSD 3-Clause license is simpler and doesn't require legal counsel to understand what is permissible.
The BSD 3-Clause license preserves two key aspects of the original caBIG license, Kerlavage explained in an NCIP blog post: it does not restrict anyone from creating derivative works based on the code, and it does not allow endorsement of derivative works using the names of contributors. While the structure of the new license is similar to the old one, the fact that it is well known in the open source community, approved by the OSI, and simpler should make it easier for bioinformatics developers to adopt it.
In addition to removing any barriers that the caBIG license may have presented, Kerlavage said that NCIP wanted to improve the mechanism by which third-party code could be contributed to the caBIG code base.
He explained that although under the previous license caBIG's code could easily be downloaded, it was difficult for third-party users to contribute enhancements they'd made back to the system because the infrastructure was created using a technology stack and build scripts that were developed internally and also because the code was stored on NCI's servers.
He said that CBIIT mulled a number of options to ease the process and decided, based on recommendations from the community, to move the software to GitHub because it is accessed frequently by bioinformatics and biomedical informatics developers.
Now, under the BSD 3-Clause license, researchers will be able to make enhancements to the code and share them with others in the community via GitHub if they choose to or they can develop and commercialize products based on caBIG, Kerlavage said. The only caveat is that those who use the code can't use the developers' names or the NCI to endorse whatever programs they develop based on the tools.
CBIIT has already begun migrating caBIG software to the NCIP GitHub channel and expects to be done by late spring or early summer, Kerlavage said. The goal is to enable the community to use the code "independent of NCI's servers, build scripts, and development environment," he said. It also requires changes to documentation, such as replacing references to the caBIG license in the code files with the BSD 3-Clause.
So far, CBIIT has moved the code for seven applications. These include the Patient Study Calendar, caArray, caIntegrator, cancer Bench to Bedside, cancer Genome-wide Association Studies, caBIG Central Clinical Participant Registry, and cancer Laboratory Information Management System. It expects to move about 50 caBIG associated programs in total, Kerlavage said.
The new license change will also apply to other software developed by CBIIT-funded efforts and the code from these projects will also be available on GitHub, Kerlavage said.