On Looking for Zebras

In lieu of a recap of the past five months, I’m going to pretend this week that I didn’t spend five months in complete radio silence.  I think that most of you are aware of the most exciting parts of my life in the past five months anyway (grad school being the largest one of those parts) so putting off the recap shouldn’t set us back too much.  Instead, this week we’re going back to the traditional “How did Alycia make a fool of herself this week?” posts.  Fair warning- this one’s a bit of a doozy (in the sense that I was a complete idiot, not that anything particularly bad happened).

One of the questions remaining to be answered in my project was, “How does this particular type of DNA damage affect the cell as a whole?”  The way we decided to try to look at this question was by treating the cells with a chemical that induces the lesions we’re interested in, and then looking at how many of the cells died after a given period of time.  The idea is that if the lesions affect the cells particularly badly, they’ll die rather than trying to live through the damage.  The method we were using to test how many cells died basically required treating the cells with the chemical that caused damage, waiting three days for the damage to take place, and then adding a second chemical that made the wells turn a different level of orange based on how many cells were still alive in that well.  All in all, it was a fairly simple procedure.

Despite the simplicity of the method, our results were not what we expected.  And beyond that, they weren’t consistent from day to day.  When your results aren’t what you expect but they’re consistent, you come to terms with the fact that you were wrong.  When your results aren’t what you expect AND they’re completely inconsistent, you go back to the drawing board and try again to see if you did something wrong the first time.  I probably ran this experiment five or six times.  Each time gave completely different results.  Something was obviously wrong.  Worse than that, since you have to wait three days for the damage to occur to visible levels, my five or six experiments took two to three weeks to get through.  Keep that in mind going forward- I spent two to three weeks getting completely bizarre, inconsistent results.

Finally, we got to the point where we decided that it had to be something I was doing- either differing amounts of chemical were being added to different sets of cells, or the cells weren’t getting into the plate correctly, but something was off.  It was decided that my grad student would observe me doing the experiment, to see if he noticed anything weird that I was doing. 

If you’ve never had someone sit behind you with the express intention of catching your mistakes, let me tell you that it’s a singularly unpleasant experience.  I’m not a person who particularly enjoys having people watch me work in the first place.  When someone is trying to watch you in order to see what you’re doing wrong, it’s even more nerve-wracking. 

Unpleasantness of my grad student watching me aside, the procedure went fine, and he didn’t see anything that I was doing wrong.  Having ruled that out, our last guess was that maybe something was wrong on the timing of the experiment- like that three days was too long to allow for damage to occur, or something to that effect.  We tested that by running a time course, adding the coloring agent at every 24 and 48 hours after treating the cells with the damaging chemical, rather than just at 72 hours like we had originally been doing.  When that experiment didn’t clear anything up, we were pretty much at our wits’ end.

Until, that is, we got into a random conversation about how the plate was loaded into the machine that reads it in order to determine the degree to which each individual well is orange.  The dish for the plate to go onto has two metal pieces in it.  I had been setting the plate on top of those metal pieces, which put the plate at a bit of an angle.  As it turns out, you’re supposed to wedge the plate into the metal pieces so that it sits flat.  Reading the plate at an angle screws up all of the values that the plate reader gives you, to the extent that (you guessed it) you might get different results each time you read a plate. 

The lesson here is probably something about not looking for zebras when you hear hoofbeats (in laymen’s terms- the simplest explanation is more often the correct one than the crazy explanation that you come up with).  It’s also something of a lesson in how easily science can go wrong.  You can read your plate wrong for three weeks.  You can put the wrong kind of chemical in an experiment (hopefully without disastrous consequences (knock on wood)).  You can forget to turn on the laser on a machine that reads things using lasers.  I’ve done all of these things, as well as a bunch of other ones that are less exciting/interesting/problematic.  Mistakes were made, and they weren’t always smart mistakes (actually, they were usually dumb mistakes :P).  The error in science, as in many things, comes from the human element.  We can reduce the error as much as possible, but errors are still mildly likely to occur.  It’s something that’s taken me a long time to realize, and I’m still working on internalizing it.  Luckily, I’m kind of excellent at making mistakes, so I’ll have plenty of instances to consider.