INSTALLATION
------------
# Requirements #
1. Operating system:
* LINUX / UNIX
2. Requisite softwares:
* MATLAB (v6.5 or later)
* Bayes Net Toolbox (BNT);
can be downloaded from http://bnt.sourceforge.net/
3. Optional software:
* NCBI BLAST suite;
can be downloaded from http://www.ncbi.nlm.nih.gov/
NOTE:
If you have not installed BLAST
locally, the way to use DBNN is:
running BLAST and constructing
your own PSSM file somewhere else
(e.g. online BLAST), and then
provide the PSSM file to DBNN.
# Compilation #
Type the following commands in the home
directory of DBNN to compile and install
the package:
make
vi rundbnn
( modify the first FOUR variables
in the rundbnn file, i.e. homedir, matlabdir,
blastdir, and dbname, according to
your own configurations and exit )
./rundbnn # the first run to check
# the installation of requisite
# softwares
If you want to launch DBNN in directories other
than the home directory of DBNN, you need to
place (or make a symbol link of) the rundbnn
program in a directory under the search path
of the system, and add the home directory
of DBNN into the search path of MATLAB.
RUN
---------
# Quick start #
The simplest way to use DBNN is typing:
./rundbnn seq.fasta output
where seq.fasta is your amino acid
sequences saved in FASTA format,
and output is the prefix of prediction
files (typically, two prediction
files will be generated:
output.fasta
and
output.raw
the former contains the predictions
in FASTA format, and the latter
contains scores of secondary structures
for each residue site). An additional
file generated by launching rundbnn
is
query.pssm
which is the PSSM file corresponding
to your seq.fasta.
By default, rundbnn reads parameters
from eight files associated with DBNN
package:
default.M1.mat
default.M2.mat
default.M3.mat
default.M4.mat
default.nn1.lin
default.nn1.sig
default.nn2.lin
default.nn2.sig
You can re-train the DBNN (see below)
and provide your own parameter files
when launching rundbnn. For example,
your parameter files are named with
prefix "newpara" (i.e. newpara.M1.mat,
... newpara.nn1.lin, ...),
you can give them to rundbnn by typing:
./rundbnn seq.fasta output newpara
If you have not installed BLAST, and
instead you have run BLAST and constructed
the PSSM file somewhere else (e.g. online),
you can use DBNN by typing:
./rundbnn yourPSSM output [ prefix-of-parameter-files ]
# Run DBN and NN separately #
1. Run DBN
type:
./dbnpred PSSM output [ prefix-of-parameter-files ]
In using dbnpred, you should always
provide the PSSM file and the prefix
of the parameter files explicitly.
Two files will be generated:
output.fasta
output.raw
where output.fasta contains the
predicted secondary structure
saved in FASTA format, and
output.raw contains the predicted
posterior probabilities distribution
for each residue.
2. Run NN
type:
./nnpred PSSM output parameters
Also two files, output.fasta and
output.raw will be generated, and
this time output.raw contains the
raw outputs of the neural network.
# Training of DBN and NN #
1. Training of DBN
type:
./dbntrain PSSM secstr.fasta parameters
where secstr.fasta is the annotation
of secondary structure for your proteins
(in FASTA format) and "parameters" is a
user-defined prefix for parameter files
that will be generated. Four files will
be generated:
parameters.M1.mat
parameters.M2.mat
parameters.M3.mat
parameters.M4.mat
2. Training of NN
type:
./nntrain PSSM secstr.fasta parameters
Four files will be generated:
parameters.nn1.lin
parameters.nn1.sig
parameters.nn2.lin
parameters.nn2.sig
=================
Xin-Qiu Yao
Jan. 7, 2007