I think you should look at this article "i am not senior at all"
How Not to Be a Bioinformatician :)
and this is quoted from a blog
"The most important 'skill' to have is PATIENCE, just enough to avoid an early death due to bashing your head against the wall when trying to get convoluted R, C, C++, Python, Perl, etc packages to run. Half your life will be spent resolving dependencies only to find that when it all finally compiles it's useless anyway. Welcome to NGS! ;-)
But seriously, nix sysadmin skills are *essential. Usually some flavor of Linux, but you may have to mess around with Solaris if you're running Sun equipment. If you're moving serious quantities of data then you're going to need decent servers (we use 64-bit, 32 CPU, 130GB RAM, 4TB onboard systems with 32TB Sun Thumpers as NAS). Even having competent dedicated sysadmins on deck will not release you from having to have solid command-line skills. Get a VPN set up or get familiar with SSH and SSH tunneling (for secure VNC for example), both can work well together and allow you to work happily remotely. Consider DynDNS if you have a dynamic IP at a remote site.
A 'scripting' language is essential. It doesn't matter what, but the bioinformatics community seems to prefer being stuck in the early 1990's with Perl while the rest of the world is moving on. As painful as Perl is when compared to something like Python, it's hard to argue against the depth of modules available.
Interpreted languages are getting better and better but you'll still need a compiled language at some point. I prefer C++ (with Boost) because it's slightly less painful than C and almost as fast when compiled with -O3.
Database skills are also essential. MySQL is a good compromise between efficiency and ease of use and nobody can argue with the price. Be sure to configure it to make the most of your systems (use block rather than file storage and tweak my-huge.cnf).
For statistics, R is popular with the bioinformatics community <strike>despite the fact that it's one of the most awful hacked-together, inconsistent, illegible, slow, lumps of crap in existence. Leveraging other peoples' R packages can sometimes be a time-saver, but if you're really serious then pay the money and use Mathematica.</strike>
Finally, there's the important aspect of visualization that often gets forgotten until it's too late. A decent charting package is worth the money. A genome browser that slots in happily with your choice of server-side setup is also advantageous.
If you're in a team of bioinformaticists and can delegate specific roles then great. Otherwise, you're expected to be a Swiss army knife."