Liz (![[info]](http://l-stat.livejournal.com/img/userinfo.gif){kind=small} fuzzyinthehead) wrote in ![[info]](http://l-stat.livejournal.com/img/community.gif){kind=small} lab_gripes,@ 2008-01-11 11:35:00 |
|
When being a user sucks
cross-posted to bipolypagangeek
I work for a contract pharmaceutical testing company. The pharma sends their product to us for testing, we test it and report the results. I'm working on a chromatographic method for a launch of a new stability batch. There is a known impurity that interferes the impurity of interest. To get around this, we have standards for both. We're supposed to run them and subtract the interfering impurity area from the other peak area for the samples. We also need to quantify the other peaks in the samples without the correction. All of these are quantified based on a standard curve using the impurity of interest.
The software I use to process the data is not capable of such things normally, but we can create custom equations when necessary. IT told me to use one of the fields they created for a similar problem. All well and good... except it doesn't work. The impurity of interest is not present in the samples and the software isn't quantifying the other peaks.
After playing around for a while, I email IT and give them a brief description of the problem. After the equivalent of "reboot and see what happens" fails, I go to their office. The main admin for this software is out with a family emergency, so I tell another person what I think. It comes out more or less like this: "The custom field you told me to use isn't working. I think it might be trying to use the curve based on that specific injection rather than the curve in general. The peak used for the curve isn't present in the samples and the correction factor is larger than any of the peaks in the samples. If the software is basing the curve on the values for that specific injection and applies the correction factor, it's going to get a negative value when it shouldn't. It's not getting anything, so I think it's returning a nul. For the injections where I don't have the factor entered in, it seems calculating correctly. Is there any way for you to check and see if the software is refering to the general saved curve or if it's refering to the curve as it applies to an individual injection?"
The guy looks a me for a second and says, "I have no idea. I understand what you're saying, but I have no clue how to check that out. It might be best to wait until Monday when Teresa comes back." Part of me feels like I could figure out the problem if I had access to the database equations, but I don't.
*sigh* I was once told by a Microsoft tech support person that I gave her the most coherent description of a problem she'd ever heard. I just wish I was as good as fixing problems as I am at pinpointing them.
cross-posted to bipolypagangeek
I work for a contract pharmaceutical testing company. The pharma sends their product to us for testing, we test it and report the results. I'm working on a chromatographic method for a launch of a new stability batch. There is a known impurity that interferes the impurity of interest. To get around this, we have standards for both. We're supposed to run them and subtract the interfering impurity area from the other peak area for the samples. We also need to quantify the other peaks in the samples without the correction. All of these are quantified based on a standard curve using the impurity of interest.
The software I use to process the data is not capable of such things normally, but we can create custom equations when necessary. IT told me to use one of the fields they created for a similar problem. All well and good... except it doesn't work. The impurity of interest is not present in the samples and the software isn't quantifying the other peaks.
After playing around for a while, I email IT and give them a brief description of the problem. After the equivalent of "reboot and see what happens" fails, I go to their office. The main admin for this software is out with a family emergency, so I tell another person what I think. It comes out more or less like this: "The custom field you told me to use isn't working. I think it might be trying to use the curve based on that specific injection rather than the curve in general. The peak used for the curve isn't present in the samples and the correction factor is larger than any of the peaks in the samples. If the software is basing the curve on the values for that specific injection and applies the correction factor, it's going to get a negative value when it shouldn't. It's not getting anything, so I think it's returning a nul. For the injections where I don't have the factor entered in, it seems calculating correctly. Is there any way for you to check and see if the software is refering to the general saved curve or if it's refering to the curve as it applies to an individual injection?"
The guy looks a me for a second and says, "I have no idea. I understand what you're saying, but I have no clue how to check that out. It might be best to wait until Monday when Teresa comes back." Part of me feels like I could figure out the problem if I had access to the database equations, but I don't.
*sigh* I was once told by a Microsoft tech support person that I gave her the most coherent description of a problem she'd ever heard. I just wish I was as good as fixing problems as I am at pinpointing them.
