You may use LC-DAD to analyse total beta carotene content of your food samples. Using Woodward Rules you can calculate the absorption maxima for 9-cis and 15-cis beta carotene which should be different and use the peak areas to calculate the ratio or percentage of each. However you may need authentic samples of each to help you in the construction of a calibration curve

Dear Mrs. Ojimelukwe, do you have standards of  9 cis and 15 cis beta carotene? If yes, I have the following suggestion to determine these isomers using a spectrophotometer. Of course, due to presence of double bonds,  these compounds absorb ultraviolet radiation. I sought  the 9-cis or 15-cis, beta carotene spectra and I found these spectra in the paper (http://www.scielo.br/pdf/cta/v21n2/7470.pdf) (Really, the paper shows the spectra to 9-cis and 13-cis) . The spectra are sparingly different. Therefore, you can determine these compounds using Classical Least Square (see: http://www.spectroscopyonline.com/classical-least-squares-part-i-mathematical-theory). However, you will need a mathematical software to resolve matrix. You can use scilab (http://www.scilab.org/). On the other hand, if you have funds, you can acquired and use matlab (http://www.mathworks.com/products/matlab/) and toolbox for multivariate calibration (Partial least Square - PLS). This is a suggestion, I don't know if will work, but there is a great probability, due to relation between theory and practical. I hope that this suggestion helps you in your research. Kind regards.

Mrs. Ojimelukwe, I think Prof Fernandes has given you good references that confirm my earlier suggestion. If you have standards then you do not need LC-MS but only LC-DAD. You are lucky since the reference give to you identifies the column, solvent and conditions for HPLC analysis

thanks for these valuable suggestions.I have the standards but the facilities are not available

Dear Philippa,

As Dr. Julio Fernandes points out, the spectra are significantly

different (linearly independent) and so can be resolved. If you have a

way to save the spectra in digital form and perform simple linear

algebra on them, you can decompose the spectrum of a mixture into the

contributions from each isomer as follows.

To make a probe spectrum for the 9 isomer, take its standard spectrum

and subtract out the component of the 9 isomer along the 15 isomer. This

gives a spectrum which is orthogonal to that of the 15 isomer so when you dot it

into the unknown spectrum it responds only to the 9 isomer. If you want

it to read out in concentration units, normalize the probe spectrum by

dividing it by its scalar(dot) product with the standard 9-isomer

(assuming that is unit concentration, say 1 uM)

To summarize, the probe spectrum for the 9-isomer is just:

R9* = R9'/(R9',R9)

where (,) signifies dot product, R9 and R15 are the standard spectra of

the two isomers, and and R9' is the un-normalized probe:

R9' = R9-[(R9,R15)/sqr(R15,R15)]*R15

When you calculate the dot product of this spectrum with the spectrum of

a mixture,it gives concentration of the 9 isomer. This follows from

Beer-Lambert law and linearity of the dot product. If the solution

contains A uM of 9- and B uM of 15-,

(R9*,(AR9 + BR15)) = A(R9*,R9) + B(R9*,R15) = A + 0

--------------------------------------------------------------------------

Likewise to make a probe for the 11 isomer, you subtract out its

component along the R9 and divided by the dot product of that with the

standard R11 spectrum:

R15' = {R15 - [(R15,R9)/sqr(R9,R9)]*R9}

R15* = R15'/(R15',R15)

-------------------------------------------------------------------------

Now what you have is a set of probe spectra that are co-orthormal with

the standard spectra: (RJ*,RK) = 1 if j=k, 0 otherwise

This is not unique, but the way we have constructed it ensures that this solution

minimizes euclidian distance (sum of squares) between the span of the RJ and the experimental spectrum, i.e. this is the classical least-squares solution

As a precaution against presence of other components or non-zero

baseline, you should always regenerate the experimental spectrum that

would result from your standard spectra combined in the concentration

you determined, and subtract to see the residual is random noise.

The precedure here actually gives the least-squares best fit of the

unknown spectrum to a linear combination of the two standards.

If baseline drift is a problem, you can also make your probe spectra

orthogonal to a constant-offset spectrum (1,1,1...). In generating the

probe spectra, subtract out also the component along (1,1,1...), which

is just the average value of the spectrum over all its

wavelength-points). Then the result will be independent of a constant

offset baseline error.

You can set this all up in a spreadsheet like Scalc or Excel (Excel has

a function that will do most of the work automatically) but if you do it

over and over again it would be worth getting a program that reads your

spec's native data and and operates on a whole series of spectra

automatically. There is a free opensource software to do this at

sourceforge: http://sourceforge.net/projects/scanedit/ , but it is not

maintained and the documentation is poor to non-existent.

You could click on "discussion" and post a question about updated

version or help with your specific problem.

 HTH,

Ed

http://sourceforge.net/projects/scanedit/

Thanks for the suggestions. Philippa

A pdf with instructions for "deconvoluting" spectra of photosynthetic pigments with the free scanedit program has been posted on the sourceforge site:

https://sourceforge.net/projects/scanedit/files/openspectralworks-1.0/doc/demo2.pdf

Thank you! Philippa