Which Operating System Do You Prefer For Bioinformatics?

Question

[Giovanni M Dall'Olio]

So, you will probably hate me for asking this question here, as there are lot of forum and blog posts on internet about it and it is also a very subjective question.

However, it may be a starting point for a good discussion, if we don't flame... Which operating system do you usually use for your work? Did you install it by yourself, and do you have administrative rights on it, or is there any IT administrator in your lab?

Answer 1

[Jeroen Van Goey]

I have no experience with them, but there are several Linux distributions out there that come preloaded with bioinformatics software.

For example:

• biobuntu which comes preloaded with amapalign, biococoa, biomode, bioperl, biopython, biosquid, blast2, boxshade, chemtool, clustalw, clustalx, dialign, easychem, fastdnaml, fastlink, garlic, gchempaint, gcubin, gff2aplot, gff2ps, EMBOSS, gmt, gromacs, hmmer, ifeffit, ifrit, imview, kalign, leksbot, libbioruby, meltinggui, mipe, molphy, mozillabiofox, mpqc/support, mummer, muscle, mustang, ncbiepcr, ncbitoolsbin, ncbitoolsx11, njplot, openbabel, perlprimer, phylip, poa, polyxmass, primer3, probcons, proda, pybliographerA, rasmol, readseq, seaview, sibsim4, sigmaalignSimple, sim4, SixpackDisplay, tcoffeeMultiple, tigrglimmerGene, treepuzzleReconstruction, treetool, treeviewxDisplays, viewmol, wise, xbs, xdrawchem and xmakemol

• Bio-Linux a fully featured, powerful, configurable and easy to maintain bioinformatics workstation. Bio-Linux provides more than 500 bioinformatics programs on an Ubuntu Linux base. There is a graphical menu for bioinformatics programs, as well as easy access to the Bio-Linux bioinformatics documentation system and sample data useful for testing programs. You can also install Bio-Linux packages to handle new generation sequence data types.

• The vision for Cloud BioLinux is to offer a base image of genome analysis resources for cloud computing platforms, such as Amazon EC2. This Science as a Service model (ScaaS) will allow us to incorporate, develop and optimize life science software as well as supporting data sets on compute clouds. This project is driven by the observation that commonly-used bioinformatics tools are hard to build and maintain, require high amounts of resources, or just too numerous to choose from.

• BioBrew is a collection of open-source applications for life scientists and an in-house project at Bioinformatics.Org. The BioBrew Roll for Rocks can be used to create Rocks/BioBrew Linux, a distribution customized for both cluster and bioinformatics computing: it automates cluster installation, includes all the HPC software a cluster enthusiast needs, and contains popular bioinformatics applications.

• Debian Med is a "Debian Pure Blend" with the aim to develop Debian into an operating system that is particularly well fit for the requirements for medical practice and research. The goal of Debian Med is a complete system for all tasks in medical care which is built completely on free software.

[Giovanni M Dall'Olio]

thanks for the answer. I think there are at least two distributions called 'biolinux' and 'bio-linux', and they are different, one is based on rpm and the other on deb. I contributed a bit to debian-med in the past, but it is more aimed at medics (there is should be also a debian-science distro).

[Hranjeev]

Cloud BioLinux and Biobuntu seems to be dead links.

[Leszek]

And why not using standard distro (ie. Ubuntu or Fedora) and just install things you really need? I setup Ubuntu every 2 years from scratch. In the end it's not that difficult once you know what you really need:)

[Shicheng Guo]

Boxshade will create and show the alignment with white-black style. Is there any way to show it as colorful style like bioedit alignment result? Thanks.

Answer 2

[Simon Cockell]

All of the 3 major platforms have their advantages, and I use all 3 practically every day. Mac OS X is my primary desktop OS, for a number of reasons, but mostly because I just seem more productive using it than any of the alternatives. All of my coding work is done over SSH on Linux (almost exclusively Ubuntu) servers. The power of Aptitude package management, and the robustness of this platform means that there really is no other choice for this kind of work. Finally I run Windows 7 on my netbook, because it is an excellent OS for that platform, and enables me to do everything I want that machine to be capable of, note-taking, blog writing, as a display machine for Powerpoint etc. It is also useful to have Internet Explorer kicking around somewhere for compatability testing.

I wouldn't consider using any machine that I didn't have admin rights on for work purposes, if I have to jump through hoops to get stuff installed, it just slows me down too much. This is another reason for using OS X for my primary desktop, it allows me to escape the University's "Common Desktop" policy for Windows PCs, which would take control of my computer out of my hands.

Answer 3

[Istvan Albert]

Tips for installing software on Max OS X:

• Install the Apple developer tools called Xcode
• Install MacPorts

You can now easily install everything from command line using the port command. List all available software

port list

Install libraries and software. etc:

port install <some-library>

[Attila Csordas]

try homebrew as unix package manager https://github.com/mxcl/homebrew

[Giovanni M Dall'Olio]

MacPorts seems to be a very useful tool, but I have the feeling that the ufficial repositories for the majority of linux distributions are bigger. Moreover, many programming languages have their own way to install new modules: easy_install for python, cpan for perl, cran for R...

[Istvan Albert]

macports is primarly for installing the necessary binary dependencies into OS X - its ability to install say python modules is just an extra feature.

[Paolo]

Porticus (http://porticus.alittledrop.com/) is a GUI for MacPorts that makes the installation of unix software as simple as using the Synaptic Frontend on debian/ubuntu.

Answer 4

[Mndoci]

Highly subjective answer of course, but I don't think you can go wrong with Linux at the backend. In general, you will find a vast variety of software, great community support, better work/$ numbers at scale. I tend to prefer debian-based distro's (Ubuntu, Debian, etc) but that's a personal choice. On the front end, I've successfully used IRIX (dating myself), Ubuntu, and OSX.

Answer 5

[Khader Shameer]

In my Master's in Bioinformatics I used RedHat 7-8 and first edition of Fedora, then for the first two years of my PhD I continued with Fedora, then moved to Ubuntu and For servers to host I used RHEL earlier and now CentOS and Ubuntu server edition. I strongly vote for Ubuntu distro for it's ease of use and strong community presence.

[Khader Shameer]

Just to update, recently moved to Mac OS X 10.6 at work. Had lot of installation issues in the beginning, but now things are on track.

[Leszek]

I have also started with RedHat, then tried Fedora. Finally, 4 years ago, I have discovered Ubuntu. And I think it's the best!

Answer 6

[Geoffjentry]

In my group, all primary computational folks (currently, a new person will be bucking this trend and going with OSX) have linux desktops that they have root access to. Most of us have a secondary machine which varies between windows & Mac OSX for various side tasks. Beyond that we have several linux based servers that people use as well (via SSH). The very few people who do any sort of computational work w/o a linux desktop ssh into these servers to do all of their real work.

[Keith Callenberg]

I have a similar experience. There are a few people in my department that stick to OS X or Windows because they like the user interface, but I notice that many tasks like compiling software from source lead to much more frustration on non-Linux systems.

Answer 7

[Nancy Parmalee]

I have a Mac Pro dual quad core that I use for everything. I use the Mac OS for day to day email, most graphics, browsing, I run a VMWare virtual Windows machine, and I do most analysis in the unix terminal. Windows runs much better on the Mac hardware, Excel is much faster in Window, and I can fileshare between the Windows and Mac systems. So far this has been sufficient for me. If I run into a situation where I need more computing power, I can ssh to a Linux server, but I haven't needed to yet. I configured the machine with 16 GB RAM, leaving slots open to add another 16 if needed. I've been happy with this.

Answer 8

[Istvan Albert]

Often people are limited to their choices by factors outside of their control. One lab that I work with requires the use of Mac computers another is using Windows mostly. Large scale computations seem to be best suited for Linux systems.

Luckily there is a migration towards unified capabilities across all platforms. Installing Cygwin on Windows allows us to tap into the power of Unix, while Linux distros have advanced graphical user interfaces like Windows and Macs.

From my own observations of non technical people, the installation of new and interdependent software packages seems to be the most difficult on Mac computers and easiest on Windows due to the computational architecture that makes all Windows computers identical.

[Jeroen Van Goey]

I would argue that the installation of new and/or interdependent software packages is easiest on Linux, because the package management system takes care of resolving all dependencies.

Answer 9

[djinn]

My tip: install Cygwin if you are using Windows

[Giovanni M Dall'Olio]

yep but with cygwin you lose all the fun of using Linux or an Unix system. Moreover, it is very complicated to install new programs (I think you have to click on the installer again) or to compile new ones. As an an alternative, you can also install a virtual machine on Windows if you really need to do that.

Answer 10

[i.sudbery]

Most stuff we do is done on a linux HPC that runs CentOS. People access this in different ways, but my preffered set up is a big windows box (Xeon, 32GB RAM) running a Xubuntu VM (which gets 20GB of that RAM).

I have SSHFS pipes set up to the disk areas on the HPC that allows me edit code in local editors, and a terminal with SSH to the cluster to run things. There is no real reason that I couldn't do this in Windows, particularly with WSL, although getting sshfs working on windows has always seem a bit of a pain. It would also probably work just as well on Mac, though I guess a Mac box that big would set me back significantly more than the windows box did (around £800).

Occasionally I want to run local RStudio or Juptyer (operating on files over SSHFS), and I find that works better on Linux than windows.

We don't have root access on the cluster, and that has occasionally been an annoyance, but to be honest conda and bioconda has eased almost all that pain, and where things arn't available in conda, I've yet to come accross something that can't be complied from source without admin rights with a little bit of effort.