I disagree with Andrew Bradbury and colleagues' suggestion that making the sequences of commercial antibodies publicly available could minimize irreproducibility in biomedical research (Nature 518, 27–29; 2015). The real solution is proper initial validation of antibodies.

In my view, the reproducibility problem is better addressed by identifying the good antibodies and the reputable companies that develop, validate and manufacture them — as astute scientists do now. Also, journals need to mandate the provision of detailed validation data, protocols and antibody sources (clone, catalogue number). Independent websites enabling the submission of antibody data and consumer feedback would also help.

The biggest investment in developing a good monoclonal antibody is the extensive work needed to validate specificity and sensitivity across all relevant applications. Unlike therapeutic antibodies, most research antibodies are not sequence-patented because the cost is too high to be recovered by sales.

Even if the practical hurdles of funding and enforcing a sequence-publishing policy could be overcome, making unpatented antibody sequences public would allow them to be widely copied, produced and sold. This would eliminate the incentive for good companies to invest in validation. It would also allow 'bad' antibody sequences to contaminate the databases.

The authors' proposal could therefore disproportionately harm the good companies, hurt the end-users it is designed to protect, and would not solve the reproducibility problem.