DROP 1.2 binary for linux kernel 2.6 with i686 32bit CPUs
Compiled and checked with CentOS 5.4
DROP 1.2 binary for linux kernel 2.6 with x86 64bit CPUs
Compiled and checked with Fedora Core 13
You will need to access the following external programs.
1. PSI-PRED (http://bioinfadmin.cs.ucl.ac.uk/downloads/psipred/psipred32.tar.gz)
2. PSI-BLAST (blastpgp - ftp://ftp.ncbi.nih.gov/toolbox/ncbi_tools/ncbi.tar.gz)
3. NR protein sequence dataset used with PSI-BLAST (ftp://ftp.ncbi.nih.gov/blast/db/FASTA/nr.gz)
4. SVM light v6.01 (http://svmlight.joachims.org/)
Download the binaries of DROP at http://www.tuat.ac.jp/~domserv/DROP.html
Extract the compressed file as
tar xvzf DROP_1.2_VERSION.tar.gz
You will find the DROP executable in DROP_1.2 directory
2. Setting DROP
Add the following environment variable to .bashrc (or .cshrc):
DROP_DIR=<DROP dir with full directory name>
if your DROP directory is /home/user/DROP, add the following sentence:
You also need to modify the following options recorded in your DROP/ref/option.txt file for setting your computer environment:
Larger CPUNum will make PSI-BLAST faster, but also decrease the performance of your PC.
CPUNum is limited by the number of your CPU Core.
Set the location of FASTA sequence database used with PSI-BLAST.
Before using DROP, you have to format this FASTA file using "formatdb" program (you can download from NCBI FTP).
Set the location of PSI-BLAST (blastpgp) and makemat executables.
Set PSI-PRED directory.
Set SVM light directory.
This is for a PC with 2 CPU cores, and when the FASTA sequence database is located in /home/user/.
blastpgp and makemat are located in /usr/share/bin/ncbi/build/;
/user/share/bin/psi_pred and /user/share/bin/svm_light are the PSI-PRED
and SVM light directories, respectively.
CPUNum = 2
NR_DB = /home/user/nr
BLAST_DIR = /usr/share/bin/ncbi/build
PSIPRED_DIR = /user/share/bin/psi_pred
SVM_DIR = /user/share/bin/svm_light
$ <DROP directory>/runDROP [target FASTA sequence with at least 1 title] [Output directory]
The output file containing the prediction results is <Output dir>prediction.txt.
The output file containing the smoothed SVM values is <Output dir>/SVM_Output/smoothed.txt.
The predicted linkers are listed in '<Output dir>prediction.txt' as follows:
TARGET_ID | LINKER_ST , LNKER_EN | LNKER_LENGTH | PEAK_VALUE | TARGET_SIZE | PEAK_POS |
 Ebina, T., Toh, H. and Kuroda, Y: DROP: an SVM domain linker
predictor trained using optimal features selected by random forest. Bioinformatics 27(4): 487-494.
 Ebina, T., Toh, H. and Kuroda, Y (2009): Loop-length dependent SVM
prediction of domain linkers for high-throughput structural proteomics. Biopolymers 92(1): 1-8.
last modification 2013. 4. 1
contact: ykuroda-AT-cc.tuat.ac.jp (replace -AT- by @)