FFAS03 2000-2012 Sanford Burnham Medical Research Institute, Godzik Lab

Fortran source code and executables for Linux

Installation:
-------------
    
1.Download ffas.tar.gz file (64 or 32 bit depending on your hardware)
  run "tar xzvf ffas.tar.gz", then a directory "ffas" will be created.

  Download sequence-pool database (uniref50s) used by FFAS to build sequence profiles (around 6.87GB)
  and put it in ffas/blast/db directory 
  (Note: current the database is unitprot at 50% non-redundant sequence identity level)

  Download disfit.ff database of FFAS calibration profiles.
  and put it in ffas/db directory

  Download the selected template databases of FFAS profiles that are current available to search against 
  in the web server,  put it in ffas/db/ directory, run "tar xzvf tdb.tar.gz", then a few template database directories 
  will be created, for example, PDB????.db.

2.Define FFAS environment variable $FFAS containing the name of the
  FFAS directory and add $FFAS/soft directory to your $PATH variable.

  For example, in .bashrc file one can add the following lines:

  export FFAS=/home/user_name/ffas
  export PATH=$PATH:/home/user_name/ffas/soft

  then open new terminal or run 'source .bashrc' your file.

3.Prepare FFAS profile(s) of your sequence(s) by first running blast.pl script to
  collect and align homologs and direct its output to profil program 
  calculating FFAS profile(s).
  
  First go to the example dir to try the following commands
  and then you can use it for other sequences in fasta format.
  
  >blast.pl < fa | tee mu | profil > ff
  
  fa - input fasta file with one or more sequences

  mu - parsed alignments from psi-blast used to calculate FFAS profile(s).
       (saving mu file is optional but recommended
        one can skip 'tee mu |' part if mu file is not needed).
       
  ff - FFAS profile(s) in text format

  Note: One can also save psi-blast results first and run profil program
  to calculate FFAS profiles. For instance:

  >blast.pl < fa > mu
  >profil < mu > ff

4.Search PDB database of FFAS profiles:

  >ffas -b ff ../PDB0614.db/fb

  -b - indicates that template database (second set of profiles) is in a binary format
  ff - query profiles (for instance profiles prepared in step 2)
  ../PDB0614.db/fb - path to a binary version of PDB profiles

5.Compare any two sets (libraries) of FFAS profiles using FFAS program:
  
  >ffas -t ffq fft > results
  
  -t - indicates that template database (second set of profiles) is in text format

  ffq - library of query profiles - one or more sequence profiles in FFAS text format
    (current version accepts libraries with up to 1200 query profiles)

  fft - library of template profiles - one or more sequence profiles in FFAS text format
    (current version accepts files with up to 100000 template profiles).

6.Large template libraries of FFAS profiles would be searched faster if they are saved
  in a binary format. To prepare a binary format of FFAS profiles run: 

  >biff ff fb

  ff - ffas profile(s) in a text format
  fb - ffas profile(s) in a binary format


7.Notes

  - lower (more negative) FFAS scores indicate stronger similarity. FFAS scores
    lower than -9.5 are expected to contain less than 3% of false positives as indicated by
    comprehensive benchmarks of known structures.

  - FFAS has optimal sensitivity for sequences longer than 50 aa and
    shorter than 600 aa.

  - see $FFAS/example/commands for an example of preparing FFAS profiles
    and comparing them with libraries of profiles.

  - when preparing large number of profiles it may be practical to split
    input fasta file into portions and calculate profiles in parallel.
    FFAS profiles in text format can be then concatenated and translated into a 
    binary format.

  - in a similar way, comparison of large libraries of profiles may be sped up
    by splitting the library of query profiles into portions and processing
    FFAS searches in parallel

  - txtff program translates FFAS profiles from binary format into text format.

Directories:
------------

ffas/source - fortran source code of ffas, profil, binff, and txtff programs

ffas/include - fortran source code of libraries of procedures used by ffas programs

ffas/soft - executables of ffas programs and scripts

ffas/blast - blast/psi-blast suite from NCBI (used to collect and align sequences needed for
             ffas profile)

ffas/blast/db - contains nr85s database - a source of homologues used to build ffas profile(s)

ffas/example/commands - a simple example of using FFAS

ffas/db/PDB0614.db - contains a library of ffas profiles of sequences from PDB database
                     (with redundancy reduced by removing identical sequences)

Compilation:
------------
64-bit executables were compiled with Intel(R) Fortran Compiler:
32-bit executables were compiled with Absoft(R) Fortran Compiler:

References:
----------
1. Rychlewski, L., Jaroszewski, L., Li, W. & Godzik, A. (2000). Protein Science 9, 232-241
2. Jaroszewski, L., Rychlewski, L. & Godzik, A. (2000). Protein Science 9, 1487-1496
3. Jaroszewski, L., Rychlewski, L., Li, Z., Li, W. & Godzik, A. (2005) Nucl. Acids Res. 33, W284-W288

Questions, comments, complaints - please contact: ffas@burnham.org
------------------------------------------------------------------