biomanycores
open-source manycores bioinformatics
biomanycores is a repository of open-source parallel
bioinformatics code
in OpenCL
(and, temporarily, in CUDA).
We would like to bridge the gap between researches in
high-performance-computing and platforms of usual bioinformaticians and biologists,
in particular by giving accesses to high-performance prototypes through Bio* frameworks.
- open-source – All projects and interfaces must be licensed under one open-source
license recognized by OSI.
- interoperability – The projects should, as far as possible, be integrated with
Open Bioinformatics Foundation Bio* frameworks (BioJava, BioPerl, Biopython...).
- portability –
The recommanded language is the OpenCL standard,
whose specification has been released in December 2008. However,
CUDA
projects can be included too.
- benchmarking –
Projects should include elements of benchmarking or research papers. Benchmarking results should be
reproductible by others.
News
- November 10: New Bioperl interface for GPU-HMMER
- Three new applications will be packaged before March.
Contact us if you are interested
- September 18: Biomanycores featured at genomeweb BioInform
- June 27: Biomanycores presented at the Bioinformatics Open Source Conference (BOSC 2009).
Biomanycores interfaces
Biomanycores interfaces are still in early stage of development. You can download the last interfaces:
All interfaces now require a working installation of CUDA and of one or several projects listed below.
You have to put the executables in the required directories.
Please consult the README files for interface-specific requirements.
Projects
SWcuda: Smith-Waterman protein alignment
pknotsRG: pseudonots of an RNA sequence
- Home (Practical Computer Science group, Universität Bielefeld, Germany)
- Authors: J. Reeder, P. Steffen, R. Giegerich
- License
- CUDA code (also works on CPU)
- Available interfaces: Biopython, BioJava, BioPerl
J. Reeder, P. Steffen, R. Giegerich,
pknotsRG: RNA pseudoknot folding including near-optimal structures and sliding windows,
Nucl. Acids. Res., Web Server Issue, 2007
cudaPWM: scan a position weight matrix against a DNA sequence
- Home (SEQUOIA group, Université Lille 1, France)
- Authors: M. Giraud, J.-S. Varré
- License: GPL
- CUDA code (also works on CPU)
- Available interfaces: Biopython, BioJava, BioPerl
- Performance : up to 77x speedup on a NVidia GTX 280 (compared to CPU)
M. Giraud, J.-S. Varré,
Parallel Position Weight Matrices Algorithms,
ISPDC 2009
GPU-HMMER: HMMER protein sequence analysis suite, cuda version of hmmsearch
Get involved
Any help that comes is welcome ! Some suggestions:
- Bio* integration: improve exisiting interfaces, include a new Bio* framework...
- New project integration: integrate a CUDA/OpenCL project, improve a CUDA/OpenCL code to make it
really usable...
