Earlier this year, my colleagues and I experienced every scientist's worst nightmare. Twelve months of experiments were deemed useless after we showed that a recommended negative control for chemically synthesized stretches of RNA (microRNA mimics), bought from a biotechnology company, was inappropriate. The sequence was too short, leading to results that were impossible to interpret, if not just wrong. Because the company didn't reveal much information about the product, we only discovered the discrepancy fortuitously after testing many microRNAs of known sequence, and observing a length-dependent activity among them.

This is the worst in a long line of incidents that we have experienced as a result of the sweeping confidentiality imposed by manufacturers of laboratory reagents, who, for the most part, do not provide full details about the contents of their chemicals, enzymes or kits. This lack of transparency forces researchers to waste time chasing information, restricts the types of experiments they can and cannot do and, most troublingly, causes them unknowingly to perform inappropriate experiments and publish misleading results.

Scant information on the myriad kits and reagents purchased by labs can lead researchers to do inappropriate experiments inadvertently. Credit: K. BROFSKY/GETTY

To try to decipher the ingredients of commercial products, my colleagues and I have tested pH and conductivity, signed confidentiality agreements to receive extra information not on the label and discarded experiments in which unknown ingredients impeded subsequent reactions. We are on first-name terms with many sympathetic scientists who work in research and development (R&D) for commercial vendors, and who occasionally whisper crucial details off the record.

This secrecy stands in stark contrast to the current practices of scientific publication. No self-respecting referee or journal would accept a research paper in which the authors relied on processes, substances or sequences that they had created themselves but did not describe in detail. Yet this is acceptable if those tools are purchased from a company. For the sake of science, everyone has to become more transparent.

The commercialization of reagents and kits is not all bad — it has undoubtedly contributed much to modern research, particularly in biological and clinical disciplines. It has enabled faster and higher-throughput experimental protocols, promoted greater uniformity and consistency between independent labs and helped to develop technologies and methodologies that would otherwise be inaccessible to individual labs.

Companies need to rethink their blanket confidentiality policies.

However, the true cost of these advances — besides the price tag — is the inability to adjust confidential methods according to experimental needs, should they exceed scenarios anticipated by the company's R&D department. For instance, molecular and cell biologists often need DNA or RNA sequences that act as negative controls — usually sequences that don't match anything in the cells being tested. Such sequence-based reagents are zealously protected by manufacturers, an understandable precaution for such an easily copied product. But in experiments that combine genomes — such as when human cells are infected with a viral genome, or when a mouse carries human DNA — a strictly human or mouse negative control might not work properly. But how would the researcher know, if they can't find out the actual sequence? Moreover, the dynamics of the industry mean that reagents (and companies) appear, merge, upgrade or disappear at short notice. These changes often take place without leaving an information trail, making it impossible to repeat or reproduce older experiments.

If we are to work towards a more rigorous and productive research community, we must change the way information about materials is disclosed. Such changes require goodwill and open-mindedness, but they are possible. For example, standards for disclosing a minimum level of information about microarray experiments, widely known as the MIAME guidelines1, have become an accepted default among microarray researchers, replacing jealous guarding of raw data and paving the way to powerful meta-analyses. A call to provide more details in experiments that use the polymerase chain reaction (PCR) in real time has, at the time of writing, been cited by more than 500 other publications2. The industry is subject to market forces, so the call for transparency should come from researchers, referees and journals voicing their concerns.

For companies, obvious and easy changes should start with full disclosure of information on discontinued products, perhaps in a centralized third-party database. This should be followed by disclosure of all non-sensitive, auxiliary information about existing products that does not need to be protected; knowing the composition of the reaction medium does not eliminate the need to purchase the enzyme, for example. On the contrary, this approach enhances products' flexibility, enabling their use in a greater variety of experiments, and thus boosting the company's potential revenue. The sequences of PCR primers are often heavily guarded by manufacturers, but some molecular biologists have suggested3 that companies could reveal alternative, less-sensitive information that would still allow scientifically rigorous analysis, such as approximate primer positioning and details about in silico validation; scientists in other fields could develop similar plans. Likewise, patent law is evolving to protect new application of genetic sequences, both man-made and naturally occurring, so at least partial disclosure of proprietary control sequences should be possible.

Transparency is not just the responsibility of manufacturers — scientists need to demand more information, and disclose everything in their methods sections. Responsible reporting by individuals who use commercial kits should spell out the principles of those kits — such as the kit version, details about the underlying biochemical reactions and all listed reagents. For software, an exact specification of version and parameters should be given. Reviewers of papers should also require this information.

Something has to be done. Because of the too-short sequence my student used as a negative control, one-quarter of her PhD was wasted. I've already seen one published paper that unknowingly reports the same artefact as a valid result. Companies need to rethink their blanket confidentiality policies: patent law safeguards their intellectual advances from financial exploitation by competitors, but it should not confound scientific rigour.