PredDBR:

Protein-DNA binding residues prediction via bagging strategy and Sequence-based Cube-Format Feature.

Pre-requisite:

• Python, Java, Pytorch
• SANN tool (https://github.com/newtonjoo/sann)
• HHblits tool (https://toolkit.tuebingen.mpg.de/#/hhblits)
• NCBI nr database (https://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/)
• Uniclust30 database (http://wwwuser.gwdg.de/~compbiol/uniclust/)

Installation:

*Download this repository PredDBR at https://github.com/YanSongCode/PredDBR/archive/PredDBR.zip first. Then, run the following command lines on linux System.

 1. unzip  -n PredDBR.zip -d ./
 2. cd PredDBR-PredDBRR
 3. cd tools/
 4. tar zxvf blast-2.2.26.tar.gz
 5. tar zxvf junh_BlastpgpSSITEOutputPARSER.tar.gz
 6. tar zxvf psipred321.tar.gz
 7. cd ..
 8. edit "config.properties" by modify the relevant path of tools or database

*The file of "config.properties" should be set as follows:

SANN_RUNNER_PATH=xx/SANN/sann/bin/sann.sh
BLASTPGP_DB_PATH=xx/nr
PSIPRED321_FOLDER_DIR=xx/psipred321/
HHBLITS_EXE_PATH=xx/hhblits
HHBLITS_DB_PATH=xx/uniclust30_2018_08
SITESEA_DB_FOLDER=xx/SiteSea
BLASTPGP_EXE_PATH=./tools/blast-2.2.26/blastpgp
BLAST_BIN_DIR=./tools/blast-2.2.26/
BLASTPGP_OUTPUT_PARSER_DIR=./tools/junh_BlastpgpSSITEOutputPARSER
PYTHON=xx/anaconda3/bin/python

Note that "xx" represent the absolute path of relevant database or tool.

Run:

java -jar PredDBR.jar [input] [input] [input] [input]

[input]Protein_name(String)  
[input]Protein_sequence(String)	
[input]Folder path to save the process files and result file
[input]{PDNA-543,PDNA-335}(String)

Note: 

   a. you should input four parameters: Protein_name, Protein_sequence, Folder path to save the process files and result file, training set;
   
      the forth parameter is a fixed parameters : "PDNA-543" or "PDNA-335"; "PDNA-543" represents the models constrcuted on dataset PDNA-543,
 
      "PDNA-335" represents the models constrcuted on dataset PDNA-335

   b. give an example: java -jar PredDBR.jar 1A02_F RRIRRERNKMAAAKSRNRRRELTDTLQAETDQLEDEKSALQTEIANLLKEKEK ./result  PDNA-543
   
   c. The outputted probability of belonging to the class of DNA-binding and non-binding of query sequences could be found at "./result/PredDBR_Prod"

Update History:

• First release 2020-12-28

References

[1] Jun Hu, Yan-Song Bai, Lin-Lin Zheng, Ning-Xin Jia, Dong-Jun Yu, Gui-Jun Zhang. Protein-DNA Binding Residue Prediction via Bagging Strategy and Sequence-based Cube-Format Feature. submitted