Linux system requirements for NGS analysis programs

[RBGSYD]

Hi,

We're seeking basic set up advice and pointers on Linux system requirements as a new lab getting into next-gen data analysis.

We may be missing some basic software or generic programs, sometimes error messages are suggesting difficulty locating gcc. We're unsure about the best place to download GNU/GCC (to ensure we can use C++ amongst other things).

Example error we've encountered:
# When configuring ghc-6.12.7
$ ...No acceptable C compiler found in $PATH

We've been attempting to install and use Bowtie 0.12.7 on Illumina sequences of whole-genomic plant DNA (7million x ~100bp reads). The aim is to separate chloroplast sequences from other DNA sequences to be able to create better de novo assemblies of the chloroplast only. Bowtie looks good, however we’re having troubles getting it up and running, most likely general programming issues.

To install Bowtie, pandoc must be installed, which requires the Haskell platform or GHC and so on. I haven’t been able to successfully configure and install any of these programs, partly because I don’t always understand the error messages. E.g.
# When attempting to install the latest Haskell platform 2011-2.0.1 for Linux (our machine is running Kernel Linux 2.6.32-122el6.x86_64)
$ …invalid configuration ‘x86_64-unkown-linux-‘: machine ‘x86_64-unknown-linux’ not recognized

We'd appreciate advice on:
-the basic system requirements to run command line programs e.g. GNU
-general advice on installing and configuring Bowtie
-other tips which are often 'assumed knowledge' but essential to be able to run command-line NGS programs effectively!

Many thanks!

[ulz_peter]

puh, that's alot to answer...

but before anyone could do that: which Linux distribution are you using? There are differences concerning file system hierarchy, configuration data etc...

Example error we've encountered:
# When configuring ghc-6.12.7
$ ...No acceptable C compiler found in $PATH

For example: to run a program you either need to specify the directory where it lies or you put it into a directory which is specified in a system variable called $PATH. (Mostly a directory in that $PATH variable is /usr/bin or /bin or something like that, this is depending on your distribution.

We'd appreciate advice on:
-the basic system requirements to run command line programs e.g. GNU
-general advice on installing and configuring Bowtie
-other tips which are often 'assumed knowledge' but essential to be able to run command-line NGS programs effectively!

Many thanks!

To execute a program you must set the executable bit and you msut have the right to execute it. You can make a file executable by:

chmod 755 yourprogram

and the $PATH thing described above

For installing various packages I would recommend you familiarize with your package installing system (depending on your distribution APT for Ubuntu, YUM for Fedora, RPM for RedHat)

Concerning the assumed knowledge: Linux and BASH knowledge is actually assumed in most bioinformatic software manuals, but there are a lot of ressources in the net.

I think for general purpose questions on Linux I recommend looking into Linux forums dedicated to your distribution, as your questions are difficult to answer and actually don't fit too nicely into that forum...

[RBGSYD]

Hi Peter,

Thanks so much for your reply. I appreciate it is a lot to ask, at this stage knowing which are the right questions to be asking is almost as important!

In reply to your comments:
- We are running Linux 6.61 (Kernel Linux 2.6.32-122el6.x86_64), using RedHat
- I am familiar with the requirements for $PATH but probably hadn't specified it correctly, so I'll try that again
- Investigating RPM for RedHat sounds like a good way to go, thanks
- Actively checking various Linux forums is certainly very helpful, although it's always great if other bioinformaticians have pointers on specific resources they found valuable in the set-up stages too.

Thanks again!

[ulz_peter]

I just saw that for Bowtie there are some precompiled binaries available. Did you try them? You wouldn't need GCC or other stuff to have it work, just download the Linux x86_64 binary make it executable and that's it ( I guess, didn't try it though). Here's the link:

http://sourceforge.net/projects/bowtie-bio/files/bowtie/0.12.7/

[RBGSYD]

It has turned out to be due to us missing some primary programs -

The precompiled binaries for Bowtie worked fine on another computer. A little comparison of installed programs showed we were missing gcc-4.4.4-13.el6.x86_64.rpm on our machine, hence it failed to run. We downloaded this program and attempted to install it but came across the error that various dependencies were missing, e.g. cloog-ppl, cpp and libcomp. After downloading a version of cloog-ppl to make up for its absence, installation failed due to "no acceptable C compiler" again.

This makes me think that when we recently rebuilt the machine we missed something vital! We'll be going back to the IT people who helped us set up the system for advice.

[ulz_peter]

if it's just the gcc missing that is easy (in case you have RPM installed):
just type

rpm -q gcc

that will give you the exact name of the package and with

rpm -i gcc-4.5.1.4.something

you may install gcc. (Of course you should substitute something with the actual package number). That would be the way if you miss other programs as well...

Hope that helps

[ulz_peter]

I forgot: you may need root rights, either by sudo or by logging in as root (in case you've got a root password) then it would look like:

sudo rpm -i gcc-....