About the PSA Server

1. About the PSA Server
   1. Table of Contents
   2. Overview
   3. Limitations
   4. Using the PSA Server
   5. Commercial Users
   6. Publication of Results

Type-1 models are for complete sequences from monomeric, single-domain, globular, water-soluble proteins in several recognized structural classes. Now, however, there is an exception, because a set of Type-1 DSMs have been included for transmembrane proteins with a beta-barrel fold like porin. In any case, the Type-1 analysis may be applied to sequences having lengths in the range from 40 to 350 residues. Type-1 DSMs are also appropriate for those subsequences of membrane-spanning proteins that are believed to extend beyond the membrane (based on a hydropathy profile). See the "Description of Type-1 DSMs" section for more information on Type-1 models.

Type-2 models, in contrast, are for either partial or complete sequences from potentially large proteins that violate one or more of the modeling assumptions embodied in Type-1 models. For example, Type-2 models are appropriate for proteins that have one or more of the following properties: (1) they are multimeric; (2) they have more than one structural domain; or (3) they are not globular or soluble (e.g., membrane-spanning proteins). Type-2 models can be applied to sequences up to 1000 residues long. See the "Description of Type-2 DSMs" section for more information on Type-2 models.

WD-repeat models are specially designed for the WD-repeat family of proteins (see the http://bmerc-www.bu.edu/wdrepeat/ pages). These models combine a particular Type-1 structural model with sequence-specific pattern information. They can be applied to sequences up to 1000 residues long. Multi-domain proteins can be handed to the server intact; the region containing the WD-repeat domain will be identified by the server automatically. See the "Description of WD-repeat DSMs" section for more information on WD-repeat models.

The PSA System determines the probable placement of secondary structural elements along the sequence. In addition, when using Type-1 models, it also determines the probable tertiary structural class of the protein. In fact, it uses knowledge of this structural class when it computes the probabilities for secondary structural elements. The output of the server can either be e-mailed to the requester, or put on the Web at a private address that is e-mailed instead. The exact nature of the output depends on the type of analysis requested.

• When using Type-1 models, the output consists of three plots which graphically summarize the probabilistic tertiary structure prediction, and a text table showing secondary structure probabilities.

• When using Type-2 models, the number of plots depends on the length of the sequence. The sequence will be broken into one to four subsequences, each of which will have a Secondary-Structure Probability plot and a Strand/Turn/Helix Probability plot. In addition, the results include the text table showing secondary structure probabilities.

• When using WD-repeat models, no secondary structure probabilities will be given. Instead, the probabilities of profile 1 and profile 2 (which may be approximated as "GHXXXVXXVXFX" and "XLASGSXDXTIKVWD", respectively, as shown at http://bmerc-www.bu.edu/wdrepeat/) will be reported if WD repeats are identified in the sequence. In addition, the aligned WD repeats will also be reported with the beta strands underlined, and a skeleton coordinate file in PDB format will be included.

The analysis algorithm is based on probabilistic Discrete State-space Models (DSMs) and optimal filtering and smoothing algorithms as described in the paper "Structural analysis based on state-space modeling" by C.M. Stultz, J.V. White, and T.F. Smith, Protein Science (1993), 2:305-314. The mathematical basis for the models and algorithms is presented in "Protein Classification by Stochastic Modeling and Optimal Filtering of Amino-Acid Sequences," by J.V. White, C.M. Stultz, and T.F. Smith, Mathematical Biosciences (1994), 119:35-75. For an extended discussion of our approach, see "Predicting Protein Structure with Probabilistic Models." by C.M. Stultz, R. Nambudripad, R.H. Lathrop, and J.V. White, pp. 447-506 in: "Protein Structural Biology in Bio-Medical Research" (1997) (Editors: N. Allewell and C. Woodward), Vol. 22B, "Advances in Molecular and Cell Biology" (Editor: E.E. Bittar), JAI Press, Greenwich.

Limitations

• You may send a suitably formatted e-mail message to the server. In this case, the results will be returned via e-mail.

• You may use our Web interface at http://bmerc-www.bu.edu/psa/request.htm, which effectively submits e-mail to the server on your behalf. In this case, you may choose to have the results returned by e-mail or by Web.

To obtain information about more detailed applications of our analysis technique, which may be useful for your specific research project, please contact James White:

James V. White
JVWhite.Com
5 Kelly Road
Cambridge MA 02139

e-mail:	jvwhite@jvwhite.com
v-mail:	617-868-3045
fax:	617-868-3372

These more detailed applications involve the use of novel DSMs that are not in the standard libraries used by the PSA server. Such DSMs can be constructed to model the specific structural hypotheses being considered for a single sequence or set of related sequences.

Publication of Results

• 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: