Experimental design

[dbernick]

How should this essential skill be taught?

Should this be a requisite to the program?

Should it be taught as content in BME200?

Should it be actively taught and evaluated as part of independent study activities?

Is this a course that the program is missing?

[AndrewUzilov]

Well, the best experience is to go through peer review and get rejected badly...

I think these things can be addressed with a journal club, such as the one we've been running at the Lowe lab. We have gotten awesome ideas relevant to our work, critically reviewed the work of others, and learned a lot about experimental design by dissecting the said (and unsaid) things in methods sections and so forth.

There was some talk of adding a student-run journal club (or clubs) as a BME seminar-like course -- whatever happened to this?

That said, there are my beliefs about what makes a journal club work well:

• It must be small. Too many people and you have more folks not participating.
• It must be focused. That is why I say "clubs", in plural. You can't have one glove that fits all. Our club works because we narrow in on things relevant to our research. It is better to have several clubs (analogous to BME 281 lab meeting "courses") where students can pick and choose what they want, rather than have some monolithic one.
• Participation is key. Ideally it should be voluntary, but alternately, you can make a journal club (or clubs) a course and require grades to be based on attendance and participation. At least rotate around the room and have each person explain a figure.
• One paper dissected intensely is much more useful than more papers done in a shallow manner. Very often the devil is in the details (more often than not).

[AndrewUzilov]

In my earlier post, I think I skipped several logical steps, so let me try to fill those in...

What kind of experiment design are we talking about? Wet lab experiments/assays? Or experiments as a whole? The former requires teaching methods and getting practical hands-on experience working in the lab. Both the former and the latter require developing critical thinking, also an understanding of systems being studied and how much is understood/understandable about them given current methods and knowledge. And some good old fashioned nitpicking.

Designing computational experiments has never been taught to me. I learned about it from working with my adviser and reading papers for those short 1-2 sentence snippets in intro or discussion where they actually motivate or justify something, as opposed to dumping data on you. As of late, it has come through talking with other people who are also designing computational experiments.

Designing wet lab ("real") experiments was taught to me in upper-level undergrad biochemistry courses focused on going over primarily literature, painstakingly, one figure at a time, then grilling you about it on exams. It built up on a whole set of intro courses where everything was presented in nice, neat textbook form, just to be shattered by contradictions in primary literature models and data taught later. Seeing the model for spliceosome assembly/function or telomere structure, change from sophomore to junior year was quite enlightening. Especially fun was pitting against each other two major papers arguing for completely different models of the same thing.

What do both of the above have in common? An emphasis on critical analysis of primary literature and discussing stuff with other people who are in a similar boat as you. That's why I think an intimate journal club is the way to go. Should it be led by a faculty members? Not necessarily - just one or two sufficiently passionate senior graduate students. It worked for us.