Table 1. Comparison of the features of DCBLAST with those of existing parallel bioinformatics software for the performance of BLAST/BLAST+ searches.
| Features | DCBLAST | HPC-BLAST | GPU-BLAST | PLAST | mpiBLAST | ScalaBLAST |
|---|---|---|---|---|---|---|
| Parallelize algorithm | MapReduce | MPIa | SIMTb | Ordered Index Seed | MPI | MPI |
| Hardware requirement | HPCc environment | HPC environment (Xeon & Xeon phi) | NVIDIAd GPUe platform | SIMDf instructions set (SSEg 2+ ) supported CPU | HPC environment | HPC environment |
| Prerequisites | Sun Grid Engine, Perl (any version 5), Path::Tiny (Perl module), Data::Dumper (Perl module), Config::Tiny (Perl module) | Intel MPI C/C++ compiler, xild (Intel linker), xiar (Intel archiving) | CUDAh 2.3+ , GCCi v4.8.2+ | GCC v4.4+ , cmake 2.8+ | mvapich2 v1.4.1 or mvapich2 v1.4.1 2 or mvapich v1.2.0 3 or OpenMPI v1.4.1 or Intel MPI C/C++ compiler | Intel C/C++ compiler, OpenMPI |
| Scalable across multithreads | Yes | Yes | Yes (GPU) | Yes | Yes | No |
| Scalable across multinodes | Yes | Yes | Not applicable | No | Yes | Yes |
| Support BLAST version | All version of NCBI-BLAST+ | All version of NCBI-BLAST+ | Not applicable | Not applicable | NCBI-BLAST 2.2.20 | NCBI-BLAST 1.1.1.1 |
| Bibliography reference | This report | Sawyer et al. (2015) | Vouzis & Sahinidis (2011) | Nguyen & Lavenier (2009) | Lin et al. (2011) | Oehmen & Baxter (2013) |
| Last update | 4/18/17 | 08/25/16 | 02/09/16 | 04/21/16 | 11/28/2012 | 08/12/13 |
| Limitations | None | Only BLASTN and BLASTP | Only BLASTP | Use only single node/similar result to NCBI-BLAST | Limited output format/Older version of BLAST | Older version of BLAST |
Notes.
a
MPI, Message Passing Interface.
b
SIMT, Single-Instruction Multiple-Thread.
c
HPC, High Performance Computing.
d
NVIDIA, Nvidia corporation.
e
GPU, Graphics Processing Units.
f
SIMD, Single Instruction Multiple Data.
g
SSE, Streaming SIMD Extensions.
h
CUDA, Compute Unified Device Architecture.
i
GCC, GNU Compiler Collection.