Gene patenting has come under legal attack.
I don't support gene patenting, for this reason: implicit in any patent is the idea that you are patenting a particular mechanism by which you can do something or solve a problem. The general idea is that if somebody else does what you're doing, they're not really going to be able to add anything else significant. If they are, their patent is probably materially different from yours, or, at the very least, licensable. The advantage of this is that you don't really stifle innovation by allowing these patents - the innovation already happened with the first discovery. Instead, you permit people to innovate knowing that if they come up with something great, they'll benefit. The patent system doesn't always work this way (and definitely needs reform - this should be a standard), but it generally tends toward that way.
The problem with a patent on a gene is that you're not patenting one mechanism of solution, you're patenting the actual problem, which means that nobody else can find innovative or different solutions to the problem. Thus, gene patenting should not be allowed.
Of course, the problem is that you need to be able to still incentivize people to LOOK for those problems.
This is where transferrable R+D tax incentives can be useful. The general idea is that if you actually do basic research into genes to make these discoveries, you should benefit. Being able to write off the cost of that research in taxes would make it much easier for companies to justify doing that type of genetic research, and thus lowers disincentives to do nonexclusive scientific research. Even if the company never makes any profit that can be taxed, the tax incentives should still be sellable to other companies - almost like a form of R+D cap and trade. That would be one effective way to limit the downside of research.
Of course, making R+D 100% writeoff-able has the problem that companies will have less incentive to perform R+D efficiently - not as terrible a thing as it could be, if it means it's more thorough, but still not ideal. The best solution would then be something along the lines of "you can write off 80% of your R+D up until you a) publish a paper or b) begin preclinical trials of a treatment attacking that pathway or c) something else that marks the finish of major research in a not-very-gameable way, etc., after which you're allowed to write off all the R+D costs that had gone into the operation that hadn't been written off yet". This keeps R+D moving quickly and focuses companies on things that actually should be effective, instead of taking wild lottery chances on things they don't think could possibly work.
Then again, the way R+D works, maybe wild lottery chances aren't such a bad thing. That exact structure would require more empirics than actually have gone into study of R+D, or at least more than I've ever found.
Edit: Even thinking practically (I understand that some groups simply won't do research if they don't get exclusivity) and from a "maximize science" perspective, not just a theory one, exclusivity should not last nearly as long as it does.
Putting the incentives thinking cap on, what I'd guess you'll see is that innovation goes up on genes that have already been discovered, but research into discovering new genes goes down. I have no idea what the net effect is because I have no idea how many people are prevented from working on gene-therapies by existing patents.
The risk you run into is that genes turn into a form of trade secret, which is how a lot of other industries protect their intellectual property (Coca Cola didn't patent its recipe, it just didn't release it. Similarly, a lot of software companies use trade secrets to protect new algorithms). That doesn't help anyone.
The other question I have reading that is that unlike the creation of commercial drugs, research into genes actually does take place reasonably effectively in a university setting, but universities have limited resources. If universities are actively not pursuing genes that private companies have already got large programs working on (in a public fashion), and would substitute over, then the innovation hit would be spread evenly among gene research and other research - in other words, you simply have less research resources going into the field as a whole, so overall innovation goes down but it's not all centered in gene research (though unsubstitutable resources can't move, so gene research would be hit more). This is a reasonably plausible outcome.
If I were a biotech company, and knowing that the value of research comes in being first to market, I'd still do gene research because it's my edge, but I wouldn't tell anyone about it until the last second I had to in order to get drugs approved by the FDA - meaning I'd functionally get exclusivity for development for the period of drug development. This has a good and a bad aspect to it - on the one hand, you'd see unnecessary duplication of research, which would be a waste of precious resources. On the other hand, companies get exclusivity directly proportional to the level of innovation they had to employ, because presumably, the higher the innovation level, the longer it'd take someone else to replicate.
Overall, I think I'd rather just see a "for the next decade, you can have a gene patent but they last 4 years, not 12/18" (I choose 4 as a rough estimate for the time it takes to commercialize a family of drugs once you've found the pathway you want... it may be 5 or 6 but that's the ballpark)..."and it's going to linearly decline to 0 over the next decade". that way others don't waste time on unnecessary research, the company still gets paid for the discovery and still gets the "first to market" benefit, but you don't have a ridiculous situation where somebody "owns" a problem and won't let anyone else solve it without paying through the nose. After a decade, no more exclusivity, because I'm guessing (praying?) that finding genes is going to be trivial in 10 years, which is why the patents get shorter and shorter over time.
However, given the binary choice between "full patent" (which is either 12 or 18 years, depending on the classification) and "no patent", the better option for innovation is probably "no patent" given how much better our computer power has been getting.