Before rolling into Bioinformatics I was active in Medical Informatics. More and more I meet people with a master in Bioinformatics, but still the majority of my peers come from different disciplines such as Biology, Computer Science, Math, Toxicology, Artificial Intelligence, or Health Sciences.
This makes me wonder, which disciplines produce good bioinformaticians.
It's only relatively recently (the last 10 years or so) that it has become possible to acquire a degree in bioinformatics. Before that, bioinformaticians fell into two general types:
biologists with an interest in programming (example: Sean Eddy)
programmers (also mathematicians, statisticians) with an interest in biology (example: Jim Kent)
I don't know that particular disciplines produce "good" bioinformaticians. In fact, I think bioinformaticians tend to be people who reject the notion of disciplines altogether. See Eddy's article "Antedisciplinary" Science for an interesting perspective.
Bioinformatics involves many disciplines, ranging from computer science/engineering to math/statistics to biophysics/molecular biology to genetics/genomics. There are almost as many flavors of bioinformatics as there are combinations of 2 or 3 of these disciplines.
A great bioinformatic(ian|ist) need not be an expert in all of these disciplines, but they will definitely cross traditional discipline boundaries. The best probably have basic, foundational knowledge in many/most of these disciplines, with focused expertise in 2 or 3 specific areas.
Many graduate bioinformatics programs are an outgrowth of a particular research department/discipline (medicine, computer science, biology, etc), and the program is tailored specifically to support research in that area. It's great that so many institutions are starting to get involved in bioinformatics research, but most programs such as this are not in a position to produce scientists like I just described. There are, however, at least a few programs that treat bioinformatics separately and truly bridge the many traditional disciplines involved with bioinformatics. IMHO, these types of programs will produce the best bioinformatics scientists.
I might be biased since I am a toxicologist myself, but I think if they come from biology they better come from general biology. I know a lot of toxicologists active in bioinformatics and I think that is not by chance. A toxicologist doesn't study one aspect of biology, he is on the look out for anything that might happen and he knows about modeling of effects and kinetics and he studies interactions.
The other thing is that whatever direction he comes from (biology or informatics/computer science) he needs a large interest in, and a great respect for the other discipline. I think that for a computer scientist one of the hardest things to grasp is that biology works in a way you would never design it yourself.
I have a bachelor degree in Biotechnologies and a master in Bioinformatics. People in my group are physicists, statisticians and computer scientists. I guess it is representative of most of the groups in bioinformatics.
As someone originally trained in Computer Science, I have noticed that I often find people with formal training in applied physics to start out with the best toolset for many bioinformatics problems. They get rigor in numerical methods, statistics. They likely have a good grasp of programming. Perhaps just as importantly, they are used to thinking about problems in terms of general laws or more formal terms rather than focusing on the particular. That is NOT to say that "only physicists make good bioinformaticians", but that there are some rigors to the training that give very helpful tools and select for people who can do serious quantitative work.
I also think good bioinformaticians can come from different fields. It mainly depends the kind of project they are working on. Some projects require more informatics skills while other are more biological and require a lower extent of computational work. Then, bioinformatician can come from different fields such as life sciences, statistics, mathematics, informatics, physics,... The common point would be a taste for inter-disciplinarity which led them to get also specialized in another field.
As David above I'm quite surprised that some of the good bioinformatician I met where physicists and became influent computational biologist. It's not the first thing you will think about but it's still interesting to see. And the reason mentioned above seems quite logical to me.
Before switching to Bioinformatics I got my Ph.D in Physics, then I worked in Computer Science data mining and recommender systems - I found that the physics background is surprisingly helpful for bioinformatics - though my biology background is severely lacking - every so often I run into biology 101 level of knowledge that realize that I don't know.
Many former colleagues of mine are statisticians and chemometricians. They all work in bioinformatics right now. Particularly the latter I have not seen mentioned yet.
Ah, no, chemometrics != chem(0)informatics. Chemometrics is the application of statistics to chemical data. You may have heard of Partial Least Squares? Discovered by a chemometrician :)
+1 for the link to Eddy's article. Great read!
I also really enjoyed the read and just sent it to the whole lab :) Thanks!