Bioinformatics at the Bench

Most bioinformaticians don’t like to do experiments. We love the biological stories, but give us a pipette and we will quickly groan. I have spent the past two years in a virology lab, mainly working on computational problems, and eternally debating if I should plunge into the pool of experimental biology. Up until now, I have only dipped my toes in a few basic experiments, but I have recently decided to stop working on purely computational projects and transition to ones that require experimental skills. Here are my reasons.

I want to learn what’s possible

Computationally, anything I can imagine is possible. Experimentally, I only have a vague notion about what the space of possibility looks like. One of the reasons why I joined an experimental lab was to fill this gap, but I have finally realized that this doesn’t happen by osmosis. I will never understand what techniques are commonly used and which ones are cutting-edge without studying the literature. Also, no amount of reading will teach me how to troubleshoot experiments that don’t work; I need to interact with people that have spent years acquiring the wisdom.

I want to work on more interesting problems

Collaborations between bioinformaticians and biologists typically work like this: 1) biologist approaches bioinformatician with a ton of data and a few questions, 2) bioinformatician spends days analyzing the data, 3) bioinformatician shares findings and waits for biologist to experimentally validate them. This rarely leads to ground-breaking results because computational analyses usually underestimate the biological complexity. Bioinformatics can be used as a hypothesis generation tool, but I find that relegating the experimental component to the end is a mistake. The ability to design and perform any experiment, and the ability to process and analyze any kind of dataset are two ridiculously useful techniques on their own, but when applied together, their usefulness compounds.