 |
 |
I have an interesting problem, which arises from work done by some of our bioscientists. It is new to me, but similar problems must have arisen many times in other companies, and I would appreciate feedback and experiences from allstst members before I try a hand-crafted solution. The problem is to identify or verify the source species of samples of snake venom. They have reference samples from several species, and the test sample should be form one of these, but it may not be. Each sample is analysed by HPLC, so the comparison is based on a profile of signal-size vs retention time. The profiles usually display about 10 clear peaks and some minor ones. Their current method of analysis is to manually identify a 'reference' peak and 5 'key' peaks. The relative retention time of each key peak to the reference peak is then calculated, and compared to the corresponding numbers for the reference profiles to give a similarity score. I would like to evaluate an alternative procedure which did not require manual selection of reference and key peaks. Are there some standard ways to calculate distances or similarities between chromatographic profiles. A simple correlation coefficient may not perform well, as the absolute retention times are somewhat variable, however, the relative retention times within a profile are more stable. Thanks for your help and interest Tim Auton -- T R Auton PhD MSc C.Math Head of Biomedical Statistics Proteus Molecular Design Ltd Beechfield House Lyme Green Business Park Macclesfield Cheshire SK11 0JL UK email: [log in to unmask] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%