Install guide

G-Blastn Install guide

1. CUDA

1). Nvidia GPU card with compute capability > 1.3
2). CUDA 5.5 or above. https://developer.nvidia.com/cuda-toolkit Install CUDA and set the xxx/cuda/bin path into PATH Environment.
Eg. Centos:
vim ~/.bashrc
Add below into the .bashrc file
export PATH=/usr/local/cuda/bin:$PATH
Save and exit edit.
source ~/.bashrc

Then /usr/local/cuda/bin path is in your PATH environment.

2. Install G-BLASTN

Get the recently version from https://github.com/OpenHero/gblastn
git clone https://github.com/OpenHero/gblastn.git
cd gblastn
chmod +x install
./install
This will:
i. Ask the user whether G-LBASTN should be added to an existing BLAST installation or whether NCBI-BLAST should be installed as well.
ii. Modify the existing NCBI-BLAST installation or download, unpack and unzip NCBI-BLAST from
ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.2.28/${ncbi_blast_version}.tar.gz
depending on what was selected by the user in (i).
iii. Compile the CUDA files.
iv. Embed G-BLASTN into the existing or downloaded NCBI-BLAST.

If the installation is successful, you should find the executable "blastn" in
"./ncbi-blast-2.2.28+-src/c++/GCC447-ReleaseMT64/bin/".

NOTE: The directory "GCC447-ReleaseMT64" might differ on your system.

Acknowledgement: The installation configuration of G-BLASTN is based on that of GPU-BLAST
(http://eudoxus.cheme.cmu.edu/gpublast/gpublast.html).

3. The G-BLASTN
   

If there is no error, you can get the binary G-BLASTN file "blastn" in directory "/ncbi-blast-2.2.28+-src/c++/GCC447-ReleaseMT64/bin/".
Then move the "blastn" file into "bin" directy by command "mv" as follows:

First go to the "/ncbi-blast-2.2.28+-src/c++/GCC447-ReleaseMT64/bin/" directory, then type

mv blastn /home/blsatn/bin/gblastn