So I have been using timeOmics and following the pipeline from the paper timeOmics: an R package for multi-omics longitudinal data integration. I was using modeling function for finding the best fit model and noticed that there were some changes with the values becoming negative. Is there any reason that is?
Yes, it’s possible, and it could be due to scaling or modeling. This is particularly the case for linear models, where a straight line (y = a + bx) will predict negative y unless a > 0 and b = 0.
So, I am currently doing a log10 transformation on my data and doing LMMS modelling. I noticed that some of the values change and the variance is small to the point of doing a z-score transformation it results in many of them having the exact same z-score values across time (Perfectly linear). Is this fine to proceed or do I have to make changes to my data?
hi @bzavala,
It sounds like the log10 transformation is ‘flattening’ the signal, and this is reflected in the lmms modelling. I guess it depends on whether you expect to extract large variation across time.
Kim-Anh