You're twenty-two... Hence, this is for a university and not a school. I have advice... but it is not my own and the author of it is equally as contreversial within science. And I don't worry about spelling or grammarz whilst on teh netz.
-".6 Don't choose your initial thesis objective" -James D. Watson, Avoid Boring (Other) People
(I will note that I believe his statements concerning race from the way I heard them sound like one the stupidist statements I've ever heard... but the advice he gave above seemed to be particularly true and so I think it should be passed on)
In other words, for a project that will take 2-3 years, find a young (if you want to work on the edge of your field) advisor willing to give you a project that is both feasible within the time frame stated and potent enough to satisfy the requirements for your project.
I will add to this,
-Don't choose your thesis objective from the results given to you by a group of friends on a casual internet web forum.
OF COURSE! If you're looking for a project that you will likely never complete within 2-3 years but just want us to throw things at you I should probobly satisfy your desires as well...
IMPOSSIBLE PROJECT O' DOOM
ABANDON ALL HOPE ALL YE WHO ENTER HERE!
Highly deceptive cheery advice: You should work on the problem of protein folding from ab initio calculations... likely using or building a program like VASP to program the quantum interactions between the molecules in an attempt to predict secondary et al, of a polypeptide sequence based solely on the sequence and environmental information. If a program is required, I highly recommend using Python, because Python is the bomb!
Then when you're done with that you can find the origin of homochirality within amino acids! Personally I feel the best location to study for this would be to investigate its possible origin from circularly polarized starlight. Of course, once again high amounts of computations will likely be involved, so once again... I SUGGEST PYTHON FTW!
!!!!