This thesis discusses our recent results using the Associative-memory, Water-mediated, Structure and Energy Model (AWSEM), an optimized, coarse-grained molecular dynamics protein folding model, to fold, bind, and predict the misfolding behavior of proteins. AWSEM is capable of performing de novo structure prediction on small alpha-helical protein domains and predict the binding interfaces of homo- and hetero-dimers. More recent work demonstrates how the misfolding behavior of tandem constructs in AWSEM is consistent with crucial aspects of ensemble and single molecule experiments on the aggregation and misfolding of these constructs. The first chapter is a review of the energy landscape theory of protein folding as it applies to the problem of protein structure prediction, and more specifically how energy landscape theory and the principle of minimal frustration can be used to optimize parameters of coarse-grained protein folding simulation models. The subsequent four chapters are reports of novel research performed with one such model.