I should have written this up on Friday evening, but seeing as I was driving to Canberra it would have been inconvenient. I did pack the Queen's laptop, but it wouldn't have worked very well in the sauna. So, this post is a tad late. Apologies.
Friday is meeting day in our department. In the morning I set up a PCR reaction, helped a grad student next door with a cell culture problem, grabbed a coffee and went to our regular two-group lab meeting.
It was quite a varied session. First up was a paper review, or rather a method critique. It was to do with two papers in Nature Medicine, published a few years apart; the first paper claimed that one-dimensional nuclear magnetic resonance (NMR) can be used to diagnose a particular cardiac condition, and the second said, actually, no it can't. The problem was that the first group had not factored in proper controls for sex and drugs (I don't know about rock and roll).
Second, one of our grad students had managed to obtain crystals of her protein, but they were too small and fragile to do much with. The good news was that she had a lot of the crystals, and that she was still at around 10 mg ml-1 protein in the drops. So the obvious experiment is to drop the concentration by a factor of five or so and seed from the crystals she has. This should give slower nucleation but larger crystals.
Her second problem was that she had been unable to obtain a saturated binding curve with the protein and its ligand. The Boss made the observation that maybe the off-rate was so slow that over the time course of the experiment equilibrium had not, in fact, been achieved. This would mean that equilibrium kinetics would not apply! The obvious way to test this is to leave the binding reaction to proceed for an arbitrarily long time (e.g. overnight) and see if it saturates.
Finally, another grad student talked about her selex experiments not working. This is related to my project so I needed to pay attention. . . there was a lively discussion about whether an experiment had actually worked, and were the bands in the control lanes of the gel background or something more sinister? The general consensus was that she has a signal to noise issue. Someone drily suggested that she under-expose the gel so that the background bands disappear, resulting in much merriment.
And here's a general point. Unfortunately not all the experiments we perform give a nice, clean yes/no result, and we do need to consider seriously the presentation and significance of the data we do obtain. When you read a paper it is as important to consider the data the authors do not show as those they do. There is nothing intrinsically wrong with publishing a 'dirty' result, as long as all controls are in place and you can show a statistically significant change.
Which is where we came in.
The meeting was over in record time, my PCR finished and I skipped the midday seminar so that I could see if it had worked . . .
