Description - Protein Secondary Structure Prediction

Description
Protein Secondary Structure Prediction
compare prediction with the crystal structure

Method :

The joint prediction method for protein secondary structure combines the result of five different prediction methods. In the original New Joint Method, the five methods are: QS: Qian-Sejnowski, Na: Nagano, PF: Ptitsyn-Finkelstein, NO: Nisikawa-Ooi, GGR: Gibrat-Garinier-Robson. They were chosen from eight different methodologies, on the basis of their performance using test structures. In the New Joint Method, Na method has been replaced by SSthread (Prediction of protein Secondary Structure using Threading). Since the different predictions are computed independently, the New joint method is well suited for parallel computing.

References :

K. Nishikawa and T. Noguchi :
"Predicting protein secondary structure based on amino acid sequence",
Methods in Enzymology, Vol.202, pp.31-44 (1991).

M. Ito, Y. Matsuo and K. Nishikawa :
"Prediction of protein secondary structure using the 3D-1D compatibility algorithm.",
Comput. Appl. Biosci. vol.13, pp.415-424 (1997).

How to use :

See 'Service status' and confirm the service is ON.
Select 'Method'.
Chou-Fasman and New Joint are available now.
- Chou-Fasman : very popular and fast to calculate but tends to be less accurate.
- New Joint : a combination of several prediction programs, therefore very heavy.
Enter PDB Entry ID code (followed by one-character chain code, if necessary) into the field 'Input PDB ID'.
Press the 'Service status' button to confirm the service status for your query.
If you want to reset the input form Press 'Reset this form'.
Then, press 'Submit' button to submit your query to the server.
Results Example

Results :

If you choose Chou-Fasman method, the prediction result contains
1. success rate of the prediction
  followd by ( number of correctly predicted structures / total number of residues )
```
PERCENT CORRECT =    44.1 %   (  30/  68)
```
2. correlation coefficient for each structure type, alpha-helix, beta-sheet and coil
  followed by ( observed, predicted as that type, correctly predicted )
```
CORRELATION IN ALPHA-BETA-COIL -.134 (  35  45  21) -.143 (   9   8   0) 0.275 (  24  15   9)
```

If you choose New Joint method, to click 'Individual Predictions' button opens the next page in which you can compare the results of five different prediction programs at a time.

  SEQUENCE  EFDVILKAAGANKVAVIKAVRGATGLGLKEAKDLVESAPAALKEGVSKDDAEALKKALEEAGAEVEVK
  HOMOLOGY  AAAAAAAAAA AAAAAAAAAA   BBBBAAAAAAAAAA            AAAAAAAAAAA AAAAAA
  NEURAL    AAAAAAAA  AAAAAAAAAA  A     A A  AAAAAAAAA       AAAAAAAAAAAAAAAAAAA
  PTIFIN     AAAAAAAA    BBBB                                 AAAAAAAA          
  SSTHREAD   AAAAAAAAA  AAAAAAAAAA     AAAAAAAAAAAAAAAAA     AAAAAAAAAAAAAAAAAA 
  ROBSON    AAAAAAAAA A AAAABAAAA AAA  AAAAAAAAAA AAAAAAA AAAAAAAAAAAAAAAAAAAAAA

Furthermore, to see accuracy in the bottom of this 'Individual Predictions' page is the hyper-link to the page for success rate of each prediction program. You also get correlation coefficient for each structure type only as to New Joint method.

  NEWJOINT        187/339 55.2%
  A               82/103  79.6%   (0.408)
  B               23/112  20.5%   (0.266)
  C               82/124  66.1%   (0.373)

Copyright Notice :

Copyright of the original New Joint Method and COMPASS program used in the SSthread by Protein Engineering Research Institute (PERI). The new version of the New Joint Method has been parallelized by Real World Computing Partnership (RWCP). All rights reserved. For further information, please contact to Tamotsu Noguchi, Real World Computing Partnership,Parallel Application Laboratory, Tsukuba Research Center, Tsukuba Mitsui Bldg., 1-6-1 Takezono, Tsukuba 305, Japan.
TEL:81(298)53-1707, FAX:81(298)53-1680, E-mail:noguchi@trc.rwcp.or.jp.

PAPIA system,