Different subfields within biostatistics

[sisyphus1]

I'm in the process of gearing up for the upcoming application season, and I've been doing some research on the different biostat programs. From looking at various departments and dissertation topics of recent grads, it seems like there are two broad subfields within biostats:

1. 'traditional' biostats (focusing on application of statistics in more traditional fields like clinical studies, epidemiology, public health)

2. 'computational' biology (application of statistics in genetics, neuroscience, biomedical sciences etc.)

Most biostat programs seemed to focus on 1 (yale, columbia, upenn, and most other schools whose programs reside in the school of public health), while there are a handful that have a focus on both (harvard, johns hopkins, university of washington, from some cursory research).

I can honestly say that I have a genuine interest in both - while 1 seems less 'sexy', I can see myself being (for example) a statistician for an NGO/Government/UN, using statistics to guide public health policies. For 2, I've taken a few courses in machine learning where I was exposed to examples of machine learning techniques used in the context of biology, and I found them absolutely fascinating. Another way to look at it would be that from a career perspective, 1 seems more relevant, while from an intellectual curiosity/research standpoint, 2 is more appealing.

Given, the above, I was wondering if people could advise on the following:

- is my broad classification correct? or too myopic?

- given that 1 and 2 are quite different, should I mention that I have an interest in both (in my personal statement) or pick one and go with it?

- if I had to stick to a field, it would probably be 2. however, 2 seems like you would need some exposure to biology, and I have none.

- aside from the schools mentioned above, which other biostatistics departments have a good deal of professors working on computational biology-type problems?

Thanks all and good luck to everyone in the upcoming application process.

[cyberwulf]

- is my broad classification correct? or too myopic?

This is pretty much spot on. Statistical genetics/computational biology is really quite distinct and shares fewer and fewer tools and concerns with more 'traditional' biostatistics. Of course, many areas of 'traditional' biostatistics (causal inference, high-dimensional models, spatial statistics, etc.) are quite new and still developing.

- given that 1 and 2 are quite different, should I mention that I have an interest in both (in my personal statement) or pick one and go with it?

I don't think it really matters. If you're interested in both, might as well mention it.

- if I had to stick to a field, it would probably be 2. however, 2 seems like you would need some exposure to biology, and I have none.

This isn't a barrier, as long as you're willing and motivated to learn. Most biostat faculty doing comp bio/statgen had little to no biology training before starting graduate school.

- aside from the schools mentioned above, which other biostatistics departments have a good deal of professors working on computational biology-type problems?

All departments ranked in the top 6 or so have good people working in both more traditional biostat and comp bio/statgen. There are some overall tendencies though, which mostly have to do with what areas the most prominent faculty work in. At the risk of oversimplifying:

Hopkins: Leans comp bio/statgen

Harvard: Balanced

Washington: Leans traditional

UNC: Leans traditional

Michigan: Leans comp bio/statgen

Minnesota: Balanced (leans Bayesian in both)