spineX RELEASE NOTES
=====================

spineX Version 2.0 -- with DSSP assignment based prediction of secondary
structure

By Eshel Faraggi, Yaoqi Zhou Lab, IUPUI, September 2011 (C)
		efaraggi@gmail.com

Please cite the following papers for spineX:
-  E. Faraggi et al., ``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)
-  E. Faraggi et al., ``Predicting continuous local structure and the effect of 
its substitution for secondary structure in fragment-free protein structure 
prediction'', Structure 17,1515-1527 (2009)
-  E. Faraggi et al., ``SPINE X: Improving protein secondary structure prediction 
by multi-step learning coupled with prediction of solvent accessible surface area 
and backbone torsion angles'', Journal of Computational Chemistry, In press (2011)


See the LICENSE file for the license terms for this software.

spineX is supplied here in source code format. It must be installed for use. The 
original program was compiled on a Linux system using Intel's ifort (freely 
available for academic use at Intel's website). The compile script given in the 
code directory assumes you have ifort or gfortran installed and that its location 
is in your path. Note: if you are using a Linux system there is some chance that 
the compiled binaries supplied with this distribution will work on your system. 
Test cautiously, at your own risk while backing up anything important.

To compile the source code, go to the code/ directory and issue the command
./compile
If no errors were reported you should be able to use spineX. Consult the 
script "compile" in the directory "code/" if errors did spring up.

You must have blastpgp installed on your system. You should set the environment 
variable spineXcodir pointing to the location of the spineX code, e.g., 
export spineXcodir=/path/
Depending on your local configuration you should also set spineXblast to point 
to the blast root directory.

The directory test/ has an example to see that your version of spineX is 
close to the official web-version. Minor differences will occur because you
are probably using a different version of PsiBLAST but your results on the test
protein chain 1URSA should not be more than about a percent different than the
original predictions available in the spXout/ directory. The trend is that the
newer your version of PsiBLAST the better will be your predictions.


MODES OF OPERATION:

(1) To run the spineX server with SKSP+ based predictions:

./spX.pl id_list_file prof/fasta_location

where

 - id_list_file contains the id of chains to predict (e.g., id1)
 
 - prof/fasta_location is the absolute path to the directory with the
   psiblast profiles with .prf suffix (e.g., id1.prf) or the fasta files
   with .fasta suffix (e.g., id1.fasta). Note, name conventions on 
   prf/fasta files are strict. The program will specifically look for 
   these names.
   
   Example for test/
   ../spX.pl list1 ./

The output from this program is labeled in the output files. The indexes 0,1 
associated with the phi and psi angles designate the number of iteration (refer 
to the spineX papers for an explanation). S_ refers to the information entropy 
calculated from the predicted probabilities. pk_ designate predicted peak 
assignments (individual prediction is +/-1 summed over the 5 instances, pkc designates 
probability to be in either peak.

(2) If you have many files and several processors on the same machine to handle them, 
use the program spXbiglist.pl to automatically split the file list and run these parts
in parallel. Type ./spXbiglist.pl at the shell prompt to get help on using this program.
Note: -a 4 is passed to blastpgp so it will try to run in parralel on four processors.

(3) To obtain the spineX prediction of secondary structure according to DSSP
assignments use the program phipsi_dssp.e. Type ./phipsi_dssp.e at the shell prompt
to get help on using this program.
