Prefilter BLAST is a novel parallel prefiltering model written for NVidia GPU. Although it could be used with other algorithms, it has been specially developed and optimized for NCBI BLAST.
- Stable version
- Working in progress (alpha development version)
- Working in progress (alpha development version)
Prefilter BLAST has been developed for NVidia GPU and its programming language CUDA. For that reason, CUDA and its dependecies must be installed. For further information, go to NVidia Developer Portal
- CUDA Runtime & Developer Toolkit (tested with CUDA 2.0)
- GCC compiler and development dependencies (glibc-dev)
- -p: NCBI BLAST type (blastp for proteins, blastn for nucleotides and blastx for metagenomes).
- -d: Database file (i.e. non-redundant NR)
- -i: Query file (i.e. AMA protein family)
- -b: Filter mode (1 for different alignments, 2 for similar alignments; default 1)
- -o: Output FASTA file
- -f: Prefilter BLAST similarity threshold (default 0.5)
- -t: NCBI BLAST threshold (default 11)
- -c: GPU card number (specially for multi-GPU cards; default 0)
- -m: NCBI BLAST substitution matrix (i.e. BLOSUM62)
- -w: NCBI BLAST word size (3 for proteins and 11 for nucleotides)
make
naudit_filter -p blastp -d db/nr.fasta.fdb -i db/Amaaa.fasta -b 2 -o salida -a 6 -t 0.7
If you have any idea for an improvement or found a bug do not hesitate to open an issue.
Prefilter BLAST is distributed under MIT License.