DISOPRED Version 3.1
Copyright 2014 D. Jones, D. Cozzetto & J. Ward. All Rights Reserved.
Here are some brief notes on using the DISOPRED 3.1 software.
Please see the LICENSE file for the license terms for the software. Basically it is free to academic users as long as you don't want to sell the software or, for example, store the results obtained with it in a database and then try to sell the database. If you do wish to sell the software or use it commercially, then please contact enquiries@ebisu.co.uk to discuss licensing terms.
DISOPRED3 represents the latest release of our successful machine-learning based approach to the detection of intrinsically disordered regions. The method was originally trained on evolutionarily conserved sequence features of disordered regions from missing residues in high-resolution X-ray structures. DISOPRED2 mainly addressed the marked class imbalance between ordered and disordered amino acids as well as the different sequence patterns associated with terminal and internal disordered regions using SVMs.
DISOPRED3 extends the previous architecture with two independent predictors of intrinsic disorder - a neural network and a nearest neighbour classifier - which were trained to identify long intrinsically disordered regions using data from the PDB and DisProt databases. The intermediate results are integrated by an additional neural network.
To provide insights into the biological roles of proteins, DISOPRED3 also predicts protein binding sites within disordered regions using a SVM that examines patterns of evolutionary sequence conservation, positional information and amino acid composition of putative disordered regions.
The program is supplied in source code form - some components must be compiled before they can be used. On a standard Unix or Linux system, DISOPRED can be compiled and installed from the src/ directory with:
make clean
make
make install
The process will place the executables in the DISOPRED bin/ directory, where the script "run_disopred.pl" expects to find them. A copy of the svm-predict program from the LIBSVM package Version 3.17 is also included for the prediction of protein binding sites within disordered regions. Full details of LIBSVM, including the licence, can be found at:
http://www.csie.ntu.edu.tw/~cjlin/libsvm/
You will additionally need to download the disored library to a directory called dso_lib. From the main disopred directory run
wget http://bioinfadmin.cs.ucl.ac.uk/downloads/DISOPRED/dso_lib.tar.gz
tar -zxvf dso_lib.tar.gz
You must also set the ENVIRONMENT variable for the DSO_LIB_PATH to the path to newly untarred dso_lib/
A simple Perl script called "run_disopred.pl" allows to predict intrinsically disordered regions and protein binding sites within them. The script assumes that the NCBI BLAST binaries and appropriate sequence databases have been installed locally. Their location is specified through the variables:
my $NCBI_DIR = "/home/bin/blast-2.2.26/bin/"; # directory where the BLAST binaries are my $SEQ_DB = "/home/uniref/uniref90"; # the path to the formatdb'ed sequence database
The NCBI executables can be obtained from ftp://ftp.ncbi.nih.gov/blast
Suitable sequence data banks are available from ftp://ftp.ncbi.nih.gov/blast/db/ and ftp://ftp.ebi.ac.uk/pub/databases/uniprot/
**** IMPORTANT NOTE ON BLAST+ ***** NCBI are encouraging users to switch over from the classic BLAST package to the new BLAST+ package. On the one hand this is a cleaner and nicer version of BLAST, but on the other hand, it omits some useful features. In particular, BLAST+ no longer offers the facility to extract more precise PSSM scores from checkpoint files in a "supported" way (i.e. using the makemat utility for this purpose).
Eventually, we will probably switch over to BLAST+ as the preferred way of searching for similar sequences, but for the time being no interface to BLAST+ is provided.
The Perl script also expects to find the directories bin/, data/ and dso_lib/ at the same path. If you need to move these directories somewhere else, please change the values of the variables with the new full paths
my $EXE_DIR = abs_path(join '/', dirname($0), "bin"); # the path of the bin directory my $DATA_DIR = abs_path(join '/', dirname($0),"data"); # the path of the data directory $ENV{DSO_LIB_PATH} = join '', abs_path("./dso_lib"), '/'; # the path of the library directory used by the nearest neighbour classifier
The script "run_disopred.pl" requires as input a text file containing one amino acid sequence for which predictions are sought. A few parameters can be tuned from inside the script, including the PSI-BLAST search options and the DISOPRED2 SVM specificity level. During the execution, a number of temporary files will be generated (e.g. PSI-BLAST output files, the PSSM file, the intermediate disordered residue prediction files, the input file to svm-predict), which are identified by concatenating the input file name, the process id of the Perl job and the numeric identifier for the host. These files are removed after the final output has been generated in the same directory as the input.
Here is the output of a successful DISOPRED run for the file examples/example.fasta:
./run_disopred.pl examples/example.fasta
Running PSI-BLAST search ...
Generating PSSM ...
Predicting disorder with DISOPRED2 ...
Running neural network classifier ...
Running nearest neighbour classifier ...
Combining disordered residue predictions ...
Predicting protein binding residues within disordered regions ...
Cleaning up ...
Finished
Disordered residue predictions in absolute-path/examples/example.diso
Protein binding disordered residue predictions in absolute-path/examples/example.pbdat
Results are saved in plain ASCII text format. Disordered region predictions are presented in tabular format with four fields on each line representing the amino acid position, the residue single letter code, the order/disorder assignment code, and the corresponding confidence level. Ordered residues are marked with dots (.) and have scores in [0.00, 0.49]; disordered residues are labelled with asterisks (*) and are scored in [0.50, 1.00].
Putative disordered protein binding sites are annotated in a similar way, with one row for each amino acid and four fields representing the sequence position, the single letter code, the assignment code, and the confidence level. Ordered residues are labelled with dots (.) and have no score associated, so the value in last field is "NA". Protein-binding disordered residues are indicated by carets (^) and their confidence scores are in [0.50, 1.00], while all other unstructured positions are tagged with dashes (-) and are scored in [0.00, 0.49].
Please cite:
Jones, D.T. and Cozzetto, D. (2014) DISOPRED3: Precise disordered region predictions with annotated protein binding acrivity, Bioinformatics