We showed nearly-deterministic relation of retention times of trinucleotides with their nucleotide composition. We modeled properties of trinucleotide molecules that actually explained most of the variability of the original model. I was invited to this study to construct models.

Paper was published in an evil journal, to get pdf please send me an email.