This is an interesting topic. Here is my two cents.
I worked in the Medical Center of a large US university for some years. I first worked in a Computational Biology research group and then in a Clinical Trials setup, where I did some bioinformatics stuff. I have a PhD in statistics, but due partly to mistakes and miscalculations on my part, wound up doing mostly coding in these jobs. While I have nothing against software work, I didn't want to spend my whole time doing that.
I can second what David Thornley said. The way I think of it is, if you aren't part of the club, people don't respect you. So for example, if you are in a biology research group, and are not a biologist, you are not in the club. They may tolerate you if they see you as a equal, say a mathematician. Unfortunately, research scientists, at least the ones I have met, don't respect computer programmers. Ironically, as DNA Coder says, these same people don't have a clue about programming, and generally make a mess. (This is assuming they actually do progamming at all, and don't farm the work off to graduate students or contract workers of some kind.) Specifically, they have no understanding of design issues or the necessity of testing, and are often happy to take the results of some software program and stick it in their publications without further ado.
The reasons for this are clear. Academia does not reward expertise in software. What is important is journal publications. Code is just of interest insofar as it generates results for publications and grant applications.
Historically, journals have not cared about software except in the marginal case where the article actually is about the software. However, this may be changing. See the Nature article linked below.
This is generally true across the research scientices, not just in biology. Disclaimer: I'm just one data point, and the people I worked with may not be the best. The clinical trials people were especially stupid. Maybe the really good scientists have better attitudes towards software engineering.
To put this in context, these days there is an increasing problem with reproducible computational research. Science needs to be reproducible, so researchers can check each others work. With the increasing importance of computation in the sciences, this means people need to be able to reproduce the results of software. However, if the software is a mess, this is not so easy. There is an interesting article in Nature on this topic. At the risk of adding more negativity to this post, some of these numbers sound very inflated. "Only 47% of scientists have a good understanding of software testing"? I think the number is probably 10% or less.
To sum up, anyone who is a good programmer and also has a good understanding of the domain area is very valuable, but don't necessarily expect the people who employ you to understand that.