Hacker News Reader

I work on computational epidemiology and complex systems research. See for example: http://www.gleamviz.org/ , http://sites.google.com/site/cxnets/research and http://cnets.indiana.edu/ . Maybe something will grab your attention.

My advice is what you don't want to hear. Go learn about another topic to postgraduate level if you really want to be at the forefront of a particular topic.

Pretty much every topic has extremely interesting and unsolved problems, but to solve them you more often than not need deep domain knowledge, or multiple supervisors (but I don't recommend that at all!).

Machine learning for drug design.

Google for a recent paper by mit grads on rankSvm for virtual screening of drug molecules.

Ask HN - Computer science interdisciplinary research