Dedicated bioinformatics and medical distributions

Between 2003 and 2005 several adapted distributions targeting bioinformatics and medicine emerged. Most of them were based on Knoppix[Knopper, 2005] (and thus indirectly on Debian). The motivation behind them was to allow a small team of not more than five people to release a product with the following features:
\begin{itemize*}
\item easy to assemble
\item tailored for the personal needs of this specific work group
\item attractive web appearance
\end{itemize*}

Due to the complexity of the task, these projects had to make compromises regarding the quality of their products and were frequently being
\begin{itemize*}
\item sloppy about policy of the underlying Debian system
\item sloppy about the licenses of the included programs
\end{itemize*}

It is hard to estimate the user base of such distributions, but it can be assumed that they were basically only used inside the institution where they were developed. As a consequence, usually not much effort was invested in setting up a complete support infrastructure: bug tracking system, active mailing lists, forums, IRC, etc.

It turned out that this approach is not sustainable for the distribution of Free Medical Software. Some of these distributions issued only a single release, others remained longer but were constantly losing manpower. Furthermore, no security updates were made available, and the distribution of general software tools stagnated at the moment at which the original fork from Debian, Knoppix or other distribution happened. Especially the latter aspect made such an approach to distribute software unacceptable for sensible medical data.

To prevent these problems, Debian Med took a different approach. The initial team was also quite small, but has grown in the eight years of its existence to more than 20 active maintainers. The primary idea is to maintain medical applications inside Debian. This automatically solves all drawbacks of the separate distributions mentioned above: security updates are provided without any specific effort of the Debian Med team. There is a huge and reliable infrastructure with a bug tracking system, several FTP servers all over the world, autobuilders and an QA team running intense tests regarding software quality (see above).

An effort similar to Debian Med exists in Fedora as FedoraMedical SIG (Special Interest Group) and in openSUSE as openSUSE Medical . Both projects share the same idea of putting medical applications inside a larger distribution instead of trying to do the work of the distributor themselves. The difference with Debian Med (besides of being different distributions) is that both teams are much smaller and thus the number of packaged applications is in both cases less than 5% of the applications available for Debian. Moreover Debian does not distinguish between a core distribution and optional extensions. Consequently, a full Debian distribution on whatever medium also automatically contains all medical applications as well.

Looking beyond Linux distributions there is also FreeBSD. The ports collection also contains a versatile collection of biological software . The remarkable fact here is not the underlying operating system (you can easily have Debian with a FreeBSD kernel with Debian 6.0) but rather the fact that the same strategy to support a specific field as in Debian Med was followed: Use a large and technically well supported system and put the specific software for special use cases into this system instead of deriving a whole new system from the existing one.

Andreas Tille 2010-12-10