Protein folding by a combination of methods

[Daniel Offutt]

Has there been any work on combining molecular dynamics simulation approaches
to protein folding with neural network methods?  I do not mean by this a
staged approach where, for example, a neural network predicts secondary
structure, and then a folding simulation program finishes the task.  I mean
a folding simulation program that has been parameterized with "weights" and
then trained to fold proteins better using a protein structure database for
training data.  If you understand what I have in mind here, and can lead me
to some relevant papers, I would much appreciate it.

Daniel Offutt