By Joan P. Schwartz, Ph.D., NINDS

What is a scientific collaboration? How can one set one up and keep it going successfully? And why do they occasionally go awry?

The NIH Guidelines for the Conduct of Research (just reprinted in a revised third edition and available from your scientific director) accentuate the positive-that "research collaborations frequently facilitate progress and generally should be encouraged." And to help eliminate the negative, the Guidelines suggest setting ground rules at the start and arranging to share reagents with collaborators outside NIH through MTAs (material transfer agreements).

But the disputes that can be generated during the course of an otherwise valuable scientific collaboration-disputes revolving around not only reagent sharing but also authorship and even mentorship-are common enough that they are among the central issues the new Ombudsman/Cooperative Resolution Center pilot project was designed to handle. 

So what is a good collaboration? The NIH Committee on Scientific Conduct and Ethics recently discussed several cases of problem-plagued collaborations and came up with what we hope are useful guidelines. First, in these days of multidisciplinary science, since almost no one is trained in all the disciplines needed to complete a study, scientific collaborations clearly make a lot of sense, both intellectually and financially. The best collaborations form between scientists with complementary expertise-for example, a molecular biologist capable of generating knock-out mice with a neuroscientist who can measure changes in the behavioral activity of those mice; or an immunologist who wants to look at the effect on T lymphocytes of engineered mutants of a virus provided by a virologist.

To work well, though, certain parameters need to be discussed and defined up front: who is going to do what and when they will do it; who will supply reagents needed for certain aspects of the study; even who will write the paper and be first author. Defining order of authorship before doing the experiments can be tricky, however, since surprise results may completely change the focus of a study and thereby dictate a change in the order. Flexibility is thus a key ingredient in any collaboration.

The cases the Ethics Committee examined have convinced us that the single most important measure in successful collaboration is keeping the lines of communication open. Communicate with your collaborators, by phone, e-mail, or even letters, frequently. Tell them what you are finding and ask what their results are. Share data as well as problems. If a collaborator outside NIH is applying for an NIH grant, or is supported by an NIH grant, the granting agency should be informed of this collaboration. You will generally be asked to prepare a letter to be submitted with such a grant application; you should ask to see the relevant parts of the application before it is submitted so that you know whether the proposal accurately represents your part of the collaboration. Although you, as an NIH employee, cannot contribute to the writing of the application, make it clear that you want to be informed when the grant is funded and when it will start. Above all, do not assume that long periods of silence indicate that your collaborator is working away and all is well. If you have not communicated with your collaborators for a year, there may no longer be a collaboration!

Bear in mind that some forms of scientific exchange do not form an appropriate basis for collaboration. The Guidelines state clearly that "individuals . . . who have assisted the research [by providing] reagents . . . should not be authors." By the same criteria, providing someone with a plasmid, or an antibody, or even a transgenic mouse, does not establish a collaboration. In line with this thinking are Public Health Service regulations that state that any reagent developed with government funds (intramurally or extramurally) must be provided to those who request it once the results have been published. Intramural scientists use MTAs when giving such reagents to colleagues at universities or other extramural sites. Such input is often acknowledged in a published study, with thanks to the suppliers of materials used in the experiments-a way to give credit without conferring authorship.

Probably the most difficult issue scientists grapple with in discussing collaborations is that of intellectual property. Is there such a thing as ownership of an idea? If there were, would anyone discuss science with anyone else? Would everyone feel that they deserved authorship or collaborator status because they had lunch with a friend, heard about new results, and suggested an interesting experiment? Conversely, are all conversations between scientists, even one-on-one, to be considered a sharing of privileged information? The members of the Ethics Committee felt overwhelmingly that no conversation between scientists could be considered "privileged and confidential" unless one of the scientists started the conversation by stating that what he or she was about to share was unpublished material and was not to be shared with others.

Many scientists believe that the constraints imposed by industry consultation and collaboration on free and open discussion of research projects are already having a deleterious effect on science. For many of us, the pleasure of doing science lies in formal and informal discussion and exchange of results and ideas with colleagues. That pleasure would be compromised or vanish entirely if each idea were fenced in as the exclusive intellectual property of one person.