![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
rebeccmeisterI ran my first circadian trial in California back on July 5, 2016. Around a month before that, a collaborator had declared a need for one of our sensitive respirometry instruments (a Li-Cor CO2 analyzer) to conduct field measurements, so the early stages of the first circadian experiment involved some scrambling to get our less sensitive respirometry setup up and running (a FoxBox). This shouldn't be a big problem, given the nature of the circadian experiments (crickets use enough oxygen and exhale enough carbon dioxide to allow relatively easy measurement with the FoxBox). But then two months later the more sensitive Li-Cor was returned to us and I was encouraged to switch back over to that setup. In case that wasn't enough fun, I had to switch back to the FoxBox again two months later, I think because someone else in the lab once again needed the more sensitive setup.
Something has plagued me ever since then: the oxygen measurements from the more sensitive setup (an OxZilla) just never seemed quite right. We'd initially dismissed this, until it became so blindingly obvious that we couldn't ignore it anymore. Unfortunately, that was at around the same time as I was trying to wrap things up in California, so I never had an opportunity to conduct a series of tests to try and figure out what was wrong.
Thankfully, here in Arizona I'm working with a respirometry guru, who has been giving me access to the PC-based software I used for the initial data collection. Conversations with J convinced me to contact the instrument manufacturers, which led to some in-depth conversation with one of their excellent technical support people. I then convinced my advisor in California to conduct a small hypothesis test, which confirmed my suspicion that something was wrong with the voltage transmission from the Oxzilla to the computer.
The only annoying thing about all of this is that it would have been tremendously helpful if only I'd recorded all three of the measurements from the OxZilla instead of just the oxygen delta (reference oxygen, sample oxygen, and oxygen delta [sample minus reference]). That would have quickly made it obvious where the problem lay.
And yet - we can do a simple algebraic correction for the voltage transmission problem. After applying this correction, as noted in the subject line, I feel SO MUCH BETTER while looking through the data. The corrected data are now way easier to work with.
And now all I need to do is re-run a whole bunch of analyses. But this is a comparatively trivial step that will just take a known amount of time, and then it will be done - hours in a day, as compared to days, weeks, months, years of my life - which is what it took to collect the data in the first place.
Something has plagued me ever since then: the oxygen measurements from the more sensitive setup (an OxZilla) just never seemed quite right. We'd initially dismissed this, until it became so blindingly obvious that we couldn't ignore it anymore. Unfortunately, that was at around the same time as I was trying to wrap things up in California, so I never had an opportunity to conduct a series of tests to try and figure out what was wrong.
Thankfully, here in Arizona I'm working with a respirometry guru, who has been giving me access to the PC-based software I used for the initial data collection. Conversations with J convinced me to contact the instrument manufacturers, which led to some in-depth conversation with one of their excellent technical support people. I then convinced my advisor in California to conduct a small hypothesis test, which confirmed my suspicion that something was wrong with the voltage transmission from the Oxzilla to the computer.
The only annoying thing about all of this is that it would have been tremendously helpful if only I'd recorded all three of the measurements from the OxZilla instead of just the oxygen delta (reference oxygen, sample oxygen, and oxygen delta [sample minus reference]). That would have quickly made it obvious where the problem lay.
And yet - we can do a simple algebraic correction for the voltage transmission problem. After applying this correction, as noted in the subject line, I feel SO MUCH BETTER while looking through the data. The corrected data are now way easier to work with.
And now all I need to do is re-run a whole bunch of analyses. But this is a comparatively trivial step that will just take a known amount of time, and then it will be done - hours in a day, as compared to days, weeks, months, years of my life - which is what it took to collect the data in the first place.
