For bioinformatics packages, open source doesn’t always mean easy installation. But Wim Van Criekinge and Luc Ducazu, former DevGen scientists and co-founders of Belgian bioinformatics consulting firm BioRebel, have launched a project in their spare time to take some of the pain out of the process.
The project, BioLinux, (www.biolinux.org) provides pre-compiled versions of popular bioinformatics packages in RPM (RedHat Package Manger) format. Ready-to-install versions of Blast, ClustalW, Phylip, BioPerl, and EMBOSS are currently available, and work on the PISE package is underway. Most of the packages are built with the Gnu compiler collection, but BioLinux has obtained permission from Intel to use parts of its C/C++ compiler, so some of the packages have been optimized to run on the Intel platform as well, Van Criekinge said.
Van Criekinge said the compiling process is often taken for granted when installing software packages. In the case of Phylip, for example, BioLinux found that a careless choice of compiler options could lead to incorrect results once the software was running. This has real-world implications for researchers who use Phylip to create phylogenetic trees that they base their published research upon, he noted. The pre-compiled versions of the software available through BioLinux can be thought of as “validated” RPMs, Van Criekinge said.
In addition to performance issues, the ready-made RPMs can also save a lot of time in the installation process. BioPerl, which Van Criekinge said can take as long as several days to install, depending on the expertise of the user, can be up and running in only 10-15 minutes from the BioLinux website. The real time savings comes in the form of dependencies, he explained — BioPerl, for example, requires the installation of at least 20 other packages.
Van Criekinge said the BioLinux team requests permission from the original developers of the software it offers through its website, so some commonly used packages, such as Fasta, are not yet available because the developers haven’t given BioLinux the okay.
BioLinux is also planning to create a reference text set for standardized Blast benchmarking, Van Criekinge said.