- most of the work to be done is the same boring stuff you do at a startup: maintain the server farm, deal with flaky libraries, tune the database. The interesting algorithm work is done by PhDs... at universities.
- the bullshit factor at bioinformatics companies is as high or higher than the trivial startups you mention. You are ultimately working on tools that people don't really know how to use and trying to sell them to people that probably don't need them. The only "real" customers are pharma companies or larger biotech firms. This makes the environment much like working for an enterprise software vendor which everyone knows isn't very fun.
- the bioinformatics companies are either in glamorous places like New Jersey, Boston suburbs or... Silicon Valley. So, you're unlikely to be able to escape from the SV scene anyway.
- at a startup these days, you at least can use slightly less painful tools like ruby and python. In bioinformatics, you are stuck with ancient Perl/CGI scripts or some VB/Java time bomb.
I spent 2 years in bioinformatics and will be forever grateful that I was given the opportunity to escape and work on silly, pointless consumer entertainment software instead.