Metabolic profile by NMR spect, 79 228 162 514 264 400 000 000 000 000 combinations possible.
Dear colleagues
I\'m very young in the field of metabolomics. We use SIMCA-P+ to calculate with PLS and OPLS the best metabolic profile (NMR-spect) that can distinguish between presence or absence of disease.
However searching the best combination within 96 metabolites is a huge manual work. If I calculate the possible combinations with 96 metabolites, 79 228 162 514 264 400 000 000 000 000 combinations are possible.
Can somebody help me with a programme that selects the combinations with the best possible sensitivity/spec for presence of disease?
Thank you
Michiel
Hello,
I am also having this kind of problem with my QSAR work. I have 37 molecules and nearly 300 descriptors to build a meaningfull QSAR model. According to my searches and reading there are many ways to select right combintaion. For examaple Statistica software has a function to choose variable form a list based on their probablistic possiblities. But some times it does not work.
\"Genetic Algorithms\" which combinies variables to select best subset is the best solution. As far as I know older versions of Statistica is capable with it.