These discrete state-space models are for either partial or complete sequences from proteins that have one, or more, of the following properties: they are multimeric, or multidomain, or nonglobular, or insoluble.

There are currently two Type-2 models: a mem_span model for membrane-spanning proteins; and a generic model for proteins having less constrained patterns of secondary structural elements. Having just this pair of Type-2 models is not as limiting as it may at first appear. There are two main reasons for this.

1. The sequence length is not a basis for analyzing proteins in a Type-2 analysis. Therefore, each Type-2 model handles sequences of any length.

2. The generic model was intentionally constructed from a minimal set of prior analysis assumptions. This model is not intended to model any particular folding class. Instead,
   • Each helix and each strand are modeled as equally probable to be either buried or exposed to surrounding water (amphipathic),
   • Each tight turn is equally probable to be either two or four residues long,
   • The average lengths of strands, helices, and loops are 4.1, 9.6, 4,3 residues, respectively. Each of the length distributions is exponential.
   • The prior probabilities and expected lengths of loops, turns, helices, and strands are consistent with a statistical analysis of proteins of known structure.

Go to:

• Submit a sequence analysis request.
• List of topics about the PSA server.
• The BMERC Home Page.

Please direct your questions and comments about these Web pages and the PSA e-mail server to: