/ Lab Management

Let's Talk About Failure

About the author:

In addition to being a scientific intern at TetraScience, Eloise Wheeler is a Senior at Harvard University studying Chemical and Physical Biology. Prior to TetraScience, Eloise was an intern at Moderna Therapeutics, Takeda Oncology, Harvard, and MIT.

Alright scientists: let’s talk about failure.

Failure and lab research go hand in hand and scientists are probably the group of people most at ease with failing. It’s great!

I first learned this as a high school intern in a lab. After nervously telling my grad student mentor that my assay didn’t work, he quipped, “Well that’s science for you! Did you learn anything? Yes? Then the experiment was worth it.”

In fact, the best experiments come when you get an unexpected result that makes you go “huh” and pushes you to try a different approach or a new way of thinking about the problem.

If you want to be successful in science, you have to be prepared for months of nothing working and hitting dead end after dead end because that is an intrinsic part of the scientific method. More parts of society need to be as comfortable as science is with making mistakes and learning from them. Books, blogs, and some excellent webcomics have covered the topic.

PhD Comics
Source: www.PhDcomics.com

Two types of research failure

Yet sometimes I think that we scientists can be so ok with failure that we tolerate things that people in other fields couldn’t abide. I like to divide failure into two types.

The first kind of failure is when you have an unexpected result. You don’t know what happened, and as you dig for an answer, you end up discovering a whole realm of literature you never considered before. Or maybe you just screwed up and will remember in the future to make sure that you’re incubating for 15, not 50, minutes. Either way, you learned something.

The second kind of failure is just plain dumb. Some examples include:

  • Your Western blot was blank because your antibody didn’t work at all, probably since you had to buy it from a sketchy company that doesn’t do proper QC.
  • The power went out and your microbial fuel cell didn’t turn back on.
  • Someone left the fridge door open so all the samples went bad.
  • You can’t trust your ELISA results because your pipets haven’t been calibrated in two years.

Infrastructure failure

Because we're so accustomed to this life of failure, there are other headaches that scientists don't consider "failure", that we also put up with all the time.

The biggest culprits are incomprehensible instrument software and arcane data analysis methods that are “just the way our lab has always done it since Joe set it up five years ago”. We often accept these problems as just a part of doing science, without pushing for solutions, even if we wouldn’t deal with it in other parts of our lives.

Maybe it’s because we are so prepared to hit dead ends in our research, so prepared for hard questions to investigate, so prepared to “just run it again”. We figure what’s the difference between repeating an experiment because of an unusual result and repeating because the plate reader broke. But I say - It doesn’t have to be this way!

Scientists deserve better, reliable infrastructure - from instruments to software, from reagents to backup systems. Infrastructure failure is not a part of the scientific process.

Almost all of the examples above have happened to me, but I will wrap up this rant with two more stories from a recent lab project I did. I was researching the microbe Shewanella oneidensis (check it out, Shewy is a really cool bug) and spent most of my time failing, in both of my fail modes.

I kept running different assays to look at the membrane potential under different metabolic conditions and kept getting weird results. When I finally dove into the literature, I found some amazing papers suggesting that S. oneidensis doesn’t do normal anaerobic respiration, instead using some new, really cool, and poorly understood mechanism.

And what did you know: my results supported that conclusion. So not only did I figure out what was going on with my experiments, I also discovered that there is a whole other metabolic pathway to investigate. Awesome! On the other hand, I couldn’t get my bacteria to grow for almost a month. It drove me insane.

Was my sterile technique wrong?
Had my media gone bad?

Nope! It turns out the incubator couldn’t control its temperature well. After I set it for 30°C and walked away, it would spike up over 40°C. S. oneidensis dies at 37°C. I was boiling my poor microbes! We had no alarms or alerts we could set, so it took a few weeks before I stumbled upon the cause of my cell growth problem. Both of these puzzles were frustrating and hard to figure out. The difference is that when I found that paper that explained all my assay results it was exciting - but when I solved the cell death mystery I was exasperated.

The best part of science is that “aha” moment! So next time you are hitting your head against a wall because nothing is working, ask yourself:

How are you failing?