LBi-DBP

Standalone packages and datasets of LBi-DBP

Pre-requisite:

Python(3), java(1.8),numpy(1.20 or higher), pytorch(1.5 or higher)
SANN software (https://github.com/newtonjoo/sann)
NCBI nr90 database (ftp://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/)
PSIPRED(psiped321) software(https://i12r-studfilesrv.informatik.tu-muenchen.de/wiki/index.php/Psipred)
HHblits tool(http://toolkit.genzentrum.lmu.de/hhblits/)
uniclust30(https://gwdu111.gwdg.de/~compbiol/uniclust/)
Linux system (suggested CentOS 7)

Installation:

First, download compressed packages in (https://github.com/wwzll123/LBi-DBP/tree/main/Standalone_Program)
second, download the FileUnion.jar in (https://github.com/wwzll123/LBi-DBP/tree/main/FileSplit)
third, union these compressed packages using FileUnion.jar by typing the following command

# [packages folder path] should be the path that contains files of Bi-DBP-standalone.zip_0 ~ Bi-DBP-standalone.zip_6.
$ java -jar FileUnion.jar [packages folder path] ./Bi-DBP-standalone.zip

fourth, uncompress the generated file of Bi-DBP-standalone.zip.
fifth, provide executable permissions for file of './jar/tools/blast-2.2.26/blastpgp'.

Set config

The files of “Config.properties” and "config.ini" should be set as follows:

Config.properties

DBS_PRED_MODEL=./jar/model/dbs/dbs.mod
BLAST_BIN_DIR=./jar/tools/blast-2.2.26
BLASTPGP_EXE_PATH=./jar/tools/blast-2.2.26/blastpgp

#Need change
BLASTPGP_DB_PATH=Absolute_Path_nr_Installtion/nr

config.ini

[PATH]

#Need change
BiP_HOME=Absolute_Path_BiP-DBP_Installtion/BiP_DBP
PSIPRED_HOME=Absolute_Path_PSIPRED_Installtion/psipred321
HHBLITS_EXE=Absolute_Path_HHsuite_Installtion/hhsuite-3.0-beta.3-Linux/bin/hhblits
HHBLITS_DB=Absolute_Path_uniclust_Installtion/uniclust30_2018_08_hhsuite/uniclust30_2018_08/uniclust30_2018_08
PYTHON_EXE=Absolute_Path_Anaconda_Installtion/anaconda2/envs/python3/bin/python (contained numpy and pytorch)
SANN_RUNNER_PATH=Absolute_Path_SANN_Installtion/SANN/sann/bin/sann.sh

Running

You should make sure your query protein sequences in the file of './workFolder/seqs.fa'.
than, enter the following command lines on Linux System.

$ cd ./Bi-DBP-standalone
$ python main.py

the predicted result will be generated in file of './workFolder/querys.jun_res'.
The first column is the prediction result, 0 means non-DBP, 1 means DBP. The second column is the probability of being predicted as a positive sample. The third column is the probability of being predicted as a negative sample.

Note

Files of .dbs_prob, .sa and .opssm are generated in './jar/example'.
The .ss and .hhm files generated by PSIPRED and HHblits are generated in './workFolder'.
If you already have these files, just put them into the corresponding folder, then the feature generator will not run. This will greatly reduce the prediction time.
If you have any question, please send email to z804908905@163.com.
All the best to you!

Reference

[1] Wen-Wu Zeng, Xuan Yu, Wenjuan Liu, Jun Hu, and Shao-Liang Peng. LBi-DBP, an accurate DNA-binding protein prediction method by lightweight interpretable BiLSTM network. Submitted.