A fundamental goal of protein design is to create artificial proteins that stably fold into well-defined three-dimensional structures,like many natural proteins does.To address this problem,we may divide it into two sub-problems.One is to define protein backbone structures that are actually designable,which means there do exist amino acid sequences that fold into them.The other is to design amino acid sequence for a given backbone structure.To address the second sub-problem,we have developed a backbone-dependent statistical model,ABACUS.It has been experimen-tally shown that for several natural backbone structures of different fold types,de novo sequences designed with ABACUS fold into desired structures with high struc-tural stability.In the meanwhile,we are applying statistical approaches to address the first sub-problem.A statistical energy model for the 3-dimensional arrangements of secondary structural elements,TetraBASE,has been developed and computation-ally tested.For backbone loops connecting secondary structural elements,statistical analysis of properties associated with designability will be discussed.