in vivo DNA synthesis (writozyme)
Imagine you have a way to synthetize arbitrary DNA inside the cell based upon an external signal (e.g., download code to the cell via light signals). Your cell is now an "interpreter" in computer parlance. You can make tweaks on the fly and see what happens. Development time decreases as you no longer need to grow a culture, you simply change the cells that are already there. Also, DNA is a great memory store - you could start logging all the various events that happen in the cell creating a "debugger". You could create DNA messages that get sent to neighbor cells for communication and further computation. You basically get a Turing machine!
Better yet, forget DNA, what if you had a ribosome that worked off an external signal rather than mRNA?
How to fix the problem that synthetic circuits tend to easily get evolved out?
- Viral transmission of desired funcion
One way to keep a population of cells running the same program is to keep virally infecting the cells with the functionality. Could we use vector-jector? If anyone tried to break free by evolving, the other cells would simply rehabilitate the vagrant cell with a painless little injection.
- Synthetic Nash equilibrium ecosystem
cell1 trades with cell2 who trades with celln in a stable equilbrium that keeps each from cheating. If you can find a very stable equilibrium that is far from other states, then you might be able to keep the system running for quite some time. Need to know the cost of a trade and the values of various goods.
In economics, the principle of comparative advantage suggests that it is better for producers to specialize and trade rather than to do everything themselves less efficiently. If there were a good way for cells to trade cargo then this specialization could occur.
- Add cell repair mechanisms to bacteria 
- Oblivious Hashing method in Software Integrity research
One way to check that the computer program you wrote is the exact one being executed by an untrusted computer is to randomly hide bits of computation in the program which compute checksums in random places. Then occasionally check the checksums with a parallel computation on a trusted machine. We probably don't have the cellular machinery necessary to implement any of this yet, but here are some pointers to the equivalent problem in computer science.
- Quorum sensing
- Conjugation - can we use the conjugation mechanism for interesting locality-based computations?
- Stochastic Computation
How might we count the number of cells in the culture? What else could we compute? Could we do image processing on an image taken by a culture of cells (shift, rotate, stretch)?
Here are some pointers on stochastic computation 
- Bacterial Perceptron
Is there a way to make a simple colony forming bacteria perform the function of a perceptron? Would be easier to culture than neurons perhaps and allow for complex computations.
Rob today mentioned that Amyris has somewhat given upon trying to compete in making biofuel, instead trying to manufacture more expensive biochemical solvents used for things other than energy.
Cell Memory - would be great to have a way to easily store, retrieve and process bits within a cell
Cells currently do this for maintaining differentation. I think there was a recent iGEM project that attempted to hijack this machinery into a custom epigenetic switch.
- Mercury Memory 
Before magnetic memory, computer bits were stored by acoustically pulsing waves down long tubes of mercury. A detector at the other end would read off the pulses and at the same time send them back to the beginning to be replayed in order to maintain the memory.
Can we create the equivalent in a culture of cells? Clearly neurons are able to do something similar. Is there a mechanism to quickly transmit signals between bacteria or yeast? Would this be useful for anything??
- Inverse Bactricity
Can you do the opposite of the 2008 Harvard project and vary chemical output based upon the electric signal. I don't think their system was reversible in this way, though could be wrong.
- siRNA 
Is there an interesting siRNA target? I don't recall any iGEM projects that dealt with this.
- Cell Rupture BioBrick
Is there a way to get the cell membrane to dismantle on cue? Could be useful for therapeutics delivery.
Any interesting pheremones to synthesize? Better mousetrap?
- Amputated E. coli
Can you synthetically get rid of the flagella in E. coli? Some strains don't have flagella.
- Synthetic organelles
Seek and Destroy
We want the ability to tag, rehabilitate or remove cells that:
a) contain a particular DNA/RNA sequence (virus infection, SNPs of known bad cells, etc)
b) don't contain a particular sequence (cancer cells that have mutated and no longer have your SNPs)
One way might be to create an E. coli or a phage with a response that is only activated in the presence of the target mRNA. The Berkeley 2006 team already has an RNA lock/key mechanism, so it seems that this should be doable if you know your target sequence up front. A phage could simply infect everything and spread that way. E. coli could perhaps use conjugation to pass a vector into all cells, but which only activates in target cells.
Caveat: this seems obvious, so either it's already been done or there is something intrisically hard about it. But, the potential applications are huge, so either way it's interesting.
DNA surface tags
Get two cells two bind if they express complimentary DNA aptamers on their surface.
Synthetically decaffeineated coffee via gene silencing
Can use VIGS (virus induced gene silencing) to turn off the various enzymes involved in caffeine production? Should be able to apply it directly to live coffee trees?
Mix red, greed, blue FP to get intermediate colors. What is a useful control mechanism? Has a color biological camera already been done? Can easily create image using 3 masks and 3 colors of light.
Powering cells with external energy
Many people want to get energy from cells (biofuels, photocells, etc). What about doing the opposite by instead feeding the cells with an external energy source (heat, laser, microwaves, electricity)? This might remove some of the issues with metabolic load and allow tremendously greater biosynthesis capability. I have no idea how to approach this yet, but it is essentially the same problem improving photosynthesis efficiency? Can you do any better than simply giving the cells abundant glucose? How often are cells ATP limited rather than by other parts of the pathway you want to optimize?
Can we create a virus that infects fruit flies and/or mosquitoes and confers immunity to malaria via siRNA? Other interesting things to do with an airborne organism? Have it emit oxytocin and vasopressin into the air to make everyone happy and loving? Would be great at parties...
Vector-Jector for Mycobacterium
From, , "Some bacteria, for example Clostridium and Mycobacterium, have no known therapeutic phages available as yet."