At our SCB Journal Club yesterday, Ryan Melvin did a talk combining methods and research. Fairly evenly divided the time between talking about different ways of doing structure prediction in case anyone was interested, and his preliminary work on looking at homology models, MD and docking and looking at structure prediction of aptamers; very difficult problems to do.
A definite take-home is to work closely with experimentalists.
Gently redacted versions of his slides are here, though a redacted recording would have been better.
Some of his slide titles are interesting: