Yep, I’m doomed.
Oct 21st 2005Wallabylaughter & nerd alert
My advisor suggested, the other day, that I experiment with digital filters to remove the fluorescence background (very broad) and the noise (very narrow) from my data (in the middle).[1] I wasn’t terribly excited, because
(1) I look nothing like a signal processing expert,
(2) I had other things planned for this week, and
(3) I’d tried it before and it didn’t really work very well.
That said, he got very excited with what I did after maybe half an hour of messing around, so I figured it was worth continuing. Yay!
Tonight, I had an idea at the gym tonight about how to automate this process… I’d found a couple papers which discussed something very similar to what I’d done by hand. They all dismissed it as very slow to find the right filter parameters, slow to actually filter any reasonable size dataset, and clearly inferior to the method they wanted to write a paper about. So I was quite excited to come up with, in about as much time as takes to write this, a way to tune the filter for a particular dataset, a way to make the process almost entirely automatic, a reasonably efficient algorithm for picking the right conditions, and a way to validate the whole process to make sure it’s possible to combine it with the other things we do to data.
To get to the title of the post: My first reaction, after putting all this together in my head, was *not* “This will make my experiment work!” It wasn’t even “This will help my labmates!” No… it was “I can write a paper about this! All by myself!” Which means that I am well and truly doomed… err, destined for academia.
But at least I’m not *quite* as bad[3] as these guys and their fake dinosaur leg.
_________________
[1] Perhaps the ideal filter would be a virtual Goldilocks to sort through each peak and say “this peak is too broad!”…. etc.[2]
[2] The danger then is that the peaks would would eat her and I’d be back where I started. Oh well.
[3] Yet. It’s probably not a good sign I’m putting foot notes here, is it.