Real-SPINE 3.0 : Solvent Accessibility and Real Value Backbone Dihedral Angle Prediction Using Guided Learning Through a Two-Layer Neural Network


This service was replced by SPINE-X


Please refer to
E. Faraggi, B. Xue, and Y. Zhou,
``Improving the prediction accuracy of residue solvent accessibility and real-value backbone torsion
angles of proteins by fast guided-learning through a two-layer neural network.'',
Proteins 74, 857-871 (2009)



Target ID (optional) :
Please paste the single-alphabet sequence below :

---

The user is assumed to have read and accepted the Terms and Conditions

For questions regarding to this server, please contact: Eshel Faraggi or Yaoqi Zhou

---

This page has been accessed times since June 02, 2008

Notes:

• Computationally the server can handle proteins of up to 10000 amino-acids. The neural network responsible for the prediction was trained with proteins of size up to 500 amino-acids. There is approximately a 1% degradation in the accuracy of the prediction for longer proteins
• The overall accuracy of this server in a ten-fold cross validation test is available from the reference above
• You can only input one sequence at a time
• Make sure to use only letters for the twenty standard amino-acids (i.e. do not use X, B, Z, and [a..z])
• The source code for Real-SPINE3.0 is available here. Instructions on how to install and use it are in the README file. You will need a working SPINE and PsiBlast (with filtered database). For consistency it is required that you use the same PsiBlast as the ones that the neural networks used for training. Per the NCBI copyright policy these files are provided here exclusively for scientific use.
• The ASA predictions for the training/testing set are available as a real value prediction file where the average and standard deviations of the ASA are calculated over five predictions, and in buried/exposed notation with a 25% cutoff (as discussed in the RealSPINE paper of 2007). Both files are offered in a single tar gzipped file. See the README file.