This article has been updated from a previous version to include details about how often the software is used and on revenue from the sales of commercial licenses.
Virginia Tech's Virginia Bioinformatics Institute this week announced that future versions of its Complex Pathway Simulator, or COPASI, software package will be released this fall under the Artistic 2.0 License, which allows third party users to add on new functionalities or bundle the tool with their own software packages and distribute the final products.
Under the terms of the new license, the source code can now be downloaded for free by commercial organizations that previously had to pay a fee to use the software. The license will not work retroactively, so users who download older versions of the software will still be subject to the restrictions of their original license.
COPASI, first published in Bioinformatics in 2006, originally operated under a dual-licensing scheme. Under one license, academic researchers could download the source code for free and use it for their research, but they weren’t allowed to redistribute their modified versions of the software. Commercial groups who wanted to use COPASI in their research were required to pay a fee. A single COPASI license was priced at $900 with discounts if customers purchased multiple licenses.
Pedro Mendes, an associate professor of biochemistry at VBI and one of COPASI’s developers, told BioInform that over time it became more “complicated” to maintain two licenses.
“I think when we all analyzed how much overhead we actually have [to do] in terms of managing two separate versions and doing two releases, we felt that it would be much easier for the team to work with just a single license and make every thing freely available,” he said.
Stefan Hoops, a computational systems biologist at VBI and a COPASI developer, added that the dual license not only prevented pharmaceutical and chemical companies from using the free version of the software even though the two versions were “identical,” but it also “imposed some restrictions in terms of what academic libraries or freely available libraries” VBI could use.
“Our license was incompatible, for example, with the General Public License, which a lot of software is released on. So we couldn’t use any libraries which [used] that [license],” he told BioInform. “Additionally, we would have to use commercial packages to create or compile our dependent libraries.”
Both Hoops and Mendes said that one of the major drawbacks of the old license scheme was that third parties weren’t allowed to redistribute the software and any new features they had built. Furthermore third parties weren’t allowed to bundle their software with COPASI into a single package which, according to Hoops, made things “cumbersome for users.”
For example, CellDesigner, a tool used to create graphical networks and biochemical models, uses COPASI to do some of its simulations. However, the developers weren’t able to distribute the tools as part of a single package because of the license restrictions, which forced users to install both applications separately and perform some operations manually to get the tools to work together.
Open Source Opens Doors …
Although commercial groups no longer have to pay to use COPASI, Hoops doesn’t see that as a financial drawback for VBI. He noted that VBI researchers and others can “sell services and applications” based on the software that will open up more commercial avenues for the tool.
“We [would] like to actually further the use of COPASI in other packages … not just for the simulation but also for building new software,” he said.
Mendes agreed that the new license will give COPASI a “bigger reach” noting that, although he didn't know the exact number of users, in the last year, COPASI was downloaded 4000 times.
[ pagebreak ]
He added that marketing COPASI was never one of the developers’ major objectives. “We are two research groups and our intention is to have software that is good for research and also to [develop and] test new algorithms,” he said.
Mendes explained that the original reason VBI released COPASI with the commercial license was simply to prevent companies from making a profit off free software without giving something back in return.
He noted that the market for the software is “small,” however, and pointed out that even if “all the pharmaceutical companies” purchased the software, the income would not offset the cost of development.
Although COPASI has generated $20,000 - $25,000 in sales of commercial licenses Mendes said that income was not a major factor in the decision.
… But the Waters are Still Murky
Earlier this year, Virginia Tech and the International Society for Computational Biology forged a partnership to make sure that GenoCAD, another tool in VBI’s arsenal, remains open source.
At the time of the partnership with ISCB, Jean Peccoud, an associate professor at VBI and one of GenoCAD’s developers, told BioInform that the agreement with ISCB was expected to smooth some of the obstacles involved in open-source licensing because it would ensure that the open-source status of GenoCAD would remain the same irrespective of any changes at the institute (BI 01/08/2010).
This week, Hoops said that he did not “see the benefit” of bringing in a third party to manage COPASI's license.
“Once we release the source code under an open source license … we cannot take that away and it is out in the open,” he said. “If we will change the license in the future that would only affect future versions.”
Peccoud noted, however, that ISCB's participation was intended to have a broader reach than simply releasing GenoCAD under an open source license, and was meant to be the bioinformatics equivalent of the Apache Software Foundation, which has developed a system of three licenses to address the original contribution of the codebase, new developers' contributions, and the end users of the software.
“The idea of making the deal with ISCB was to say, ‘Look, you could be the Apache Foundation for bioinformatics’ and we could start with GenoCAD and then once we are done with that add other software unto the portfolio managed by ISCB,” he told BioInform.
According to Peccoud, one of the problems with the traditional open source model in academic settings is that a developer who creates software often doesn’t have the authority to choose the license under which the software operates. That responsibility lies with the technology-transfer office of the developer’s institution.
“The best way of [getting an open source license] is to actually work with the tech-transfer office and convince them open source [is better],” he said.
[ pagebreak ]
Peccoud said working with VT’s tech-transfer office was his original intent for GenoCAD, but pointed out that VT could change the license at anytime. Furthermore, he noted that most “mature open-source software” is usually managed by a foundation that deals with all the licensing legalities.
“In many cases, it’s an organization that represents the community, [so] then the community is protected because the software is in the hands of an organization that represents the interests of the community and not a particular party,” he said.
He used Cytoscape as an example. Although the tool was originally developed at the Institute for Systems Biology, its developers formed a separate not-for-profit organization to deal with licensing issues since they had multiple developers in different universities.
The model adopted by the Apache Foundation is a little different, Peccoud said, since the foundation uses a consistent system of licenses to manage its software portfolio.
Peccoud noted that since the ISCB deal was inked in January, there has been little or no “follow through” from the society, which he speculates may be because the organization is “stretched for resources.”
The agreement is “kind of in a limbo right now,” he said.
Peccoud said that he and others are looking at other organizations that could take on an Apache Foundation-like role for bioinformatics software, and plan to make an announcement soon.
He declined to elaborate, and ISCB could not be reached in time for a comment.
With COPASI, “the story is more complicated” Peccoud said, since half of the grant used to develop the tool was awarded to Germany’s European Media Laboratory, and each partner has different policies for handling and licensing open-source software.
He expressed some surprise at the choice of the Artistic License for COPASI, stating that it’s “not something that’s very common for software.”
But, according to Mendes, the Artistic License was chosen in part because it “fit fairly well with the research groups’ intentions,” and because it has legal precedence.
The Artistic License was created by the developers of Perl programming language and is used in some software packages such as the Parrot virtual machine.
New License, New Version
For the first version of COPASI under the new license, which is expected to be released this fall, Hoops said that the developers plan to add several new features based on feedback from a user forum, workshops, and tutorials. He would not comment on the new features, however, saying the team hadn’t finalized what they would be.
Mendes confirmed that the final list of features is still a work in progress, but did say that, among others, the developers are working on incorporating “various types of non-linear dynamics analysis such as bifurcation analysis” and that they are parallelizing the software to run “potentially four times faster” on machines with multiple cores.