Repeating Experiments

One of the big things in scientific research (and probably all research), is reproducibility. I can show that the Blah Gene is directly correlated with cancer, but if another research group somewhere else cannot reproduce my results, then it means nothing.

Except generally, reproduction experiments never get published, because they aren’t sexy, and since not publishing is akin to never getting more funding, no one does it. However, that is a complaint for another time.  Today’s complaint is regarding unpublished data! And that is why all the names of genes and proteins and students have been changed to protect identities and such.

Years ago, a student we shall call Elle made a mutant of my SuperCoolGene (SCG), and did Yeast Two-Hybrid (Y2H). Very briefly, Y2H tests to see if two proteins directly bind. You put two plasmids into yeast and then plate the yeast on selective media plates, and wait to see if the yeast grows. If it grows, the proteins bind. If they do not grow, they don’t bind.

Elle tested Mutant SCG against Wildtype SCG (for reasons) and also Boop, a known interactor of SCG. Both SCG and this mutant SCG bind Boop, all is well, we move on.

Elle graduates on to better and brighter things. Years later, I want to retest Mutant SCG because other students in the lab are finding this particular mutation in SCG is doing cool stuff in various tissues. I was teaching myself how to do Y2H so I could test SCG against novel candidates, so I decided I would teach new student kid how to do Y2H. The kid was working with Mutant SCG as well as insulin signaling, but in the past, Elle never tested Mutant SCG with Insulin-Related-Thing (IRT). Also, Elle’s plasmid containing Mutant SCG was long gone.

A quick note on the Gateway Cloning System. It is amazing. Invitrogen is amazing. You add attB ends onto a gene using PCR, then clone your gene into a pDONR plasmid, and then clone that into whatever plasmid you want. In our case, pDest22 or pDest32, which are specific for Invitrogen’s Two-Hybrid system, which is also amazing and expensive.

Elle left us with many DNA samples and plasmids, and pDONR-Mutant was one. The kid, on the other hand, had been using the QuikChange kit (Thanks Agilent!) to general several other mutations similar to the one in question today. So we had a competition to see who could generate pDest32-Mutant faster: me with Gateway or him with QuikChange. Considering I had other things also going on, he won.

So we ago ahead and transform yeast and do the Y2H and behold, Mutant SCG does NOT bind Boop, and it does not bind IRT. We try it a second time. Still not binding these things. Now, if it didn’t bind only IRT, that could be a new finding and be neat. But it is supposed to bind Boop according to Elle’s previous finding.

We sequence the mutations in both. Both have the correct mutation. I go back and finish creating Elle’s version of pDest32-Mutant SCG, and we plated that experiment this week (because while it is more convenient for me to plate everything myself, the kid wanted to help and the other students he normal helps out were MIA for finals or other miscellaneous things, so he was bored).  Tomorrow is actually the 3 day mark where I take pictures of plates, but I checked them today because I was bored.

All the controls look good. Elle’s Mutant SCG binds IRT and Boop. The Kid’s Mutant SCG does not bind IRT or Boop.

(╯°□°)╯︵ ┻━┻

This is frustrating on so many levels. There must be something wrong with one of the plasmids, and if not, then there must be a reason why Elle’s Mutant will bind, but the kid’s mutant will not bind.

So back to sequencing, and this time, more than just the mutation itself. Also, running an agarose gel to see if maybe there is some drastic change in plasmid size that is rendering the kid’s plasmid nonfunctional. Another possibility is that his plasmid did not transform into yeast properly, so I need to figure out how to isolate plasmids from yeast (they make a kit for that, right?), or some other way to see if the yeast took up pDest22 but not 32, or vice versa (considering the yeast grows on the selective media without the additional chemical, at least one plasmid must have been taken up).

This is concerning, especially since reproducibility is a concern in the research world.  If I can’t reproduce a result someone from my own lab generated, then how is some lab elsewhere supposed to reproduce it? Is the current test a fluke, or was the original test a fluke?

As much as I love to troubleshoot why experiments went wrong, I wish this would get cleaned up…because  this isn’t even my project.

┬──┬◡ノ(° -°ノ)


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s