Hey guys,
So I am planning an experiment where I will check for reactions between two different compounds with distinct UV spectral curves. I was wondering how I might go about doing stats on the results?
My instinct is to just do T. tests on the peak values over time, but that seems extremely crude. I'm sure there must be better ways of doing things? What do you think? Has anyone got any experience or suggestions with this?
You don't give any real information. Your question is mostly chemistry not statistics. Start with a good book on instrumental chemical analysis. Once you know the the expected chemistry you can start looking into statistics. Your problem is quite like the analysis of a nucleic acid in solution with a cell lysate containing some protein.that in my day was done by comparing A260 with A280 lambda max for the two main peaks. See an appropriate analytical chemist, who should be able to advise you. When you have the analysis down then ask about the statistics. Best wishes David Booth
Are you comparing curves that look something like the following ?
https://media.springernature.com/full/springer-static/image/art%3A10.1038%2Fs41598-021-83322-w/MediaObjects/41598_2021_83322_Fig1_HTML.png
Target testing Factor Analysis could be useful and relevant to your study. You can have a look at the mathematical framework and experimental applications of the method in spectroscopy.
You can have a look at the book: Factor Analysis in Chemistry, by Edmund Malinowski
One starting point would be to think of how you would assess things visually and how you would measure this this quantitatively. Perhaps more simply, you could choose two fixed frequencies to record and compare things using a@ cross plot
David A. Jones thanks! I guess it is hard to say how one could even assess it visually without having the results. For example, what I would expect is that if they are in some way reacting together to form a larger molecule of some kind I would expect to see the two initial peaks drop and a new peak to emerge shifted to the longer wavelengths as they would have higher molecular weights and very crudely would be expected to absorb at higher wavelengths (I think). But I don't know for sure how they would react, so hard to say!
Okay so the cross plot is a good idea so maybe one of these?
1. Cross plot + linear regression between the two initial peaks to see change in relative conc over time (hypothesised decrease)
2. A three-way plot between two initial peaks + emerging peak (if it is sufficiently distinct from the two initial peaks)
3. Between one of the initial peaks and the emerging peaks
Sal Mangiafico yes, in fact, I am familiar with that paper and Clothianidin and Thiamethoxam would be two of the compounds that I would be getting spectra for.
Are you measuring several instances of the same compound ? Or just one curve per compound ?
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers