A threshold requirement for patent protection is the existence of patent-eligible subject matter. Many people are surprised and puzzled to learn that it is possible to patent DNA sequences. The patent statute says that you can patent any new and useful process, machine, manufacture or composition of matter. Although DNA sequences might seem like they should fail the test of novelty because they are found in nature, patent applicants avoid this obstacle by claiming DNA sequences in forms that only exist through human intervention, such as isolated and purified cDNA molecules, recombinant vectors, and transformed host cells that contain recombinant vectors. Such patents are not infringed by the DNA in the cells of our bodies, and therefore the DNA in our cells does not defeat the novelty of these claims.

So far in the US, DNA sequences have been patented as compositions of matter, a characterization that highlights their material existence as molecules. But in the modern area of high-throughput DNA sequencing, much of the value of newly identified DNA sequences resides in databases of information rather than in tangible molecules. It is not clear how to draft patent claims that will permit patent owners capture of the value of this information. One strategy that some patent applicants have tried is to claim the DNA sequences stored in machine-readable form. This is a very interesting development, and raises questions that have not been resolved by the established practice of allowing patents on DNA molecules. Can patents be used to protect data? Just a few years ago the answer to this question would have been an easy “no”. Today, following judicial expansions in the patent system to accommodate information technology, it is not so clear. Some decisions involving patents for software and other computer-implemented inventions arguably open the door to patents claiming data in machine-readable form, although I hope that the USPTO and the courts will reject these claims. They may well decide that pending claims on DNA sequence stored in computer-readable medium simply prove too much. One paradox that these claims raise is that the USPTO posts the text of issued patents on a website, as do other patent offices [12]; that website would arguably infringe patent claims on DNA sequences stored in machine-readable medium, suggesting a need to rethink this issue.

Patent rights are not well adapted to protecting information, particularly information about the natural world. Given that independent discovery of such information is quite likely to happen without the efforts of any particular patent holder, excessive protection of such information as intellectual property may slow down subsequent research more than it promotes the original data collection.

The legal category that is doing the most work right now in determining which DNA sequences may be patented is utility. In order to be patentable, a new invention must be useful. This requirement is written into the US constitution, which authorizes Congress to issue patents to promote progress in the useful arts [13], as well as in the patent statute: “Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter…may obtain a patent therefore…” [14], but it is not entirely clear what this limitation means, or what function it serves in limiting what may be patented. The courts have sometimes treated utility as a subject matter limitation that restricts the patent system to applied technology, as opposed to abstract knowledge, or to “the world of commerce” rather than “the realm of philosophy” [15], although that distinction is ambiguous in this setting. The utility requirement is also embedded in disclosure requirements that compel patent applicants to teach the public not only how to make their inventions, but also how to use them: the US Code requires that a patent application include a disclosure that provides “a written description of the invention, and of the manner and process of making and using it” in terms that “enable any person skilled in the art…to make and use the same” [16]. Another possible way of understanding the function of the utility requirement is that it serves as a timing device that determines whether an invention is ripe for patent protection, and when it would be better left outside the patent system until more work is done to understand it.

The meaning of the utility requirement has been much disputed in biotechnology and chemistry, where structures are typically discovered or built before their functions are well understood. At what point is such a discovery ripe for patent protection? The USPTO seems to vacillate between very strict enforcement of this requirement and very lax enforcement, with no clear policy basis for choosing one approach over another. The current trend in the USPTO is toward a relatively strict standard, at least as applied to DNA sequences [17], but it is not clear whether the courts will back them up if patent applicants appeal from rejections of their patent claims.

Disclosure requirements have had growing importance in US patent law as applied to genomics discoveries. In order to get a patent an inventor must provide a written description of the invention, an enabling disclosure of how to make and use the invention, and disclosure of the best mode contemplated by the inventor of practicing the invention [16]. An applicant may not update this disclosure after filing a patent application, but must get it right the first time or else lose the benefit of original filing date (or even risk invalidation of an issued patent). Apart from informing the public, patent disclosures also demonstrate what it is that the applicant has done that justifies issuing a patent. The Court of Appeals for the Federal Circuit, which is the primary appellate court in patent matters in the US, has used the disclosure requirements in recent years to limit the scope of rights available to those who identify new DNA sequences [18]. Indeed, as a general matter, that court has been generous with patent applicants in matters pertaining to patent validity, but very strict with patent owners in determining the scope of patent rights. Some US practitioners summarize these positions in the aphorism that the Federal Circuit will not rest until it has held every US patent valid, but not infringed.

The disclosure requirement constrains the timing of patent applications: one cannot get a patent until one discloses the sequence. At an earlier point in the history of gene patenting, the effect was to deny patents to inventors who had an enabling methodology and a viable strategy for cloning a gene, but had not yet actually done so and therefore could not disclose the DNA sequence [19]. This approach also constrains the scope of claims and prevents patent applicants from getting broad patents that cover many variations on a sequence on the basis of a disclosure that does not explain how to sort through those possible many variations and determine which ones will work [20]. This is an interesting area to monitor in genomics and it is a limitation that is better grounded in recent appellate case law than utility.

A final requirement for patent protection that should play a larger role in genomics than it has done so far is the non-obviousness requirement, or “inventive step” as it is known in European law. In order to be patentable, an invention must be non-obvious to a person of ordinary skill in the field, in view of the prior art or a previously disclosed knowledge in the field [21]. The meaning of that standard is much contested in the patent system. In a formulation that resonated with the courts in the past, the non-obviousness standard distinguishes the unpatentable work of the ordinary mechanic from the patentable advances of more insightful inventors. One way of understanding this distinction is that you do not need patents to bring about mundane advances that are within easy reach of the average person working in the field, but only for the non-obvious advances that require something beyond routine work. But the drafters of the current version of the US patent statute, who were uncomfortable with past judicial efforts to distinguish between patentable and unpatentable work in terms of the nature of the inventive work, added to the definition of non-obviousness, a sentence that discourages inquiries into how the inventor went about making the invention: “Patentability shall not be negativated by the manner in which the invention was made” [21]. Instead of asking how the invention was made, one should look at the results, and ask whether the invention itself is non-obvious in light of the prior art.

This approach has arguably led to the patenting of trivial work in genomics. In the era of high-throughput DNA sequencing, the discovery of new DNA sequences has become highly routine and mechanical. But the fact that new DNA sequences are identified through the work of robots and computers that could be put to work by people of ordinary skill in this highly accomplished field does not seem to preclude them from being patented. For now, the Federal Circuit seems to think that the language of the statute precludes consideration of the routineness of the sequencing method in determining whether a sequence may be patented. Perhaps as the PTO sorts through patent applications that are currently pending, the law will start to shift toward a more robust non-obviousness standard in the future.