For biology -- especially the genomics side -- Perl was absolutely dominant 10 years ago. It's been gradually edged out by other things (substantially Python and R) since then. Desire for better statistical tools may be a driver here, but it's interesting that it's led to a switch of languages rather than building more stats tools for Perl
I would say that Perl is still dominant in computational genomics. About a year ago I wrote some code to do some basic comparative analysis at the genome level alongside some more gene focused phylogeny efforts. Since the whole thing was quite simple it wasn't a problem to write the overall structure, but getting the bioinformatics Python libraries working was a pain. Even if the code tiself was Python the API was littered with various Perl and Bash idioms and there was a major bug in at least one of the tree building methods (had to patch it myself). Calls to subprocesses within the libraries (a heavy part of most bioinformatics work, where there are thousands of stand alone command line programs) would often fail without warning. The state of the art in this area is definitely behind Ruby and far behind Perl.
However, Python is still my main language and I am very happy with that. The numpy/scipy stack and everything building around that is incomparable to anything in Perl of Ruby. Also for the last month or so I have been doing work in the IPython notebook. It has really helped my productivity and documentation efforts. And it looks really cool.