I now have my experimental results, at long last. And, well, I have to say that I seem to have something.
This batch was set up with fifteen different chemical structures, and I was looking for the same effect in each run. For each structure, there was an experiment that should have shown the effect I'm after, producing a new product, then a control (without a key ingredient) which should have shown little or nothing, and another control (with the key ingredient present, but with another compound added that should have blocked it from doing anything.)
I thought at first that I had fallen completely on my face, because as I looked over the first five or ten structures, I saw the same discouraging pattern. Next to nothing in the blank control runs, which was fine. But the numbers from the corresponding key-ingredient experiments were identically low, which was the inescapable sign of nothing going on. I'd also seen some where the experimental run showed something, but the corresponding blank showed the same exact levels - again showing that my experimental conditions weren't changing anything, but that I just had a high background rate of reaction.
I was using strong language by the time I got to number twelve, which showed a pretty high value in the experimental run. I looked over to the blank run, expecting to see the same levels, another high-background dud - but it really was a blank this time. Almost nothing there. The experimental run was at least fifty times higher. I held my breath as I looked at the second "inactivating control" run, and there it was -it went right back to the blank value, as it should if my hypothesis was correct. I had set up all the experiments in duplicate, and this morning I got the repeat data, which matched the first set very tightly. It appears to be real.
And to go along with this, these experiments also included the best candidate from my first attempts, the one that got me excited about running these follow-ups in the first place. It had made a lot of product again, although just three or four times the high background rate in the blank reaction, as it had been before. But had the inactivating control experiment knocked it back down to those values anyway? It had.
And both that one and the new winner are structurally quite similar, and they're the only two of that exact class that I've run. It's becoming increasingly clear to me that I finally have something that works, after three years of on-and-off attempts. I hardly know what to do with myself.
Well, that's not quite true. I have another variation ready to go, with ten or fifteen new structures of a different type, and I have time to incorporate what I've learned from this run before they go off. It's time to see just how far this stuff can be pushed.