Chromatography Forum

Dear Andy and others:

Predictions of retention time for HILIC (and until recently RP) were based on amino acid content. Relative retention correlated with the sum retention coefficients for individual amino acids.Retention time prediction in RP has recently taken some major strides forward with a sequence-specific algorithm.

How readily could an algorithm be constructed for peptide tR prediction in ERLIC mode? Would an analagous approach using the sum of AA retention coefficients (determined numerically from retention data for peptide standards) be useful here?

In principle, construction of a similar prediction program for HILIC should be straightforward. In either HILIC or RP (reversed phase), a large percentage of the amino acid residues will make some degree of contribution to retention, and the degree will vary with the amino acid involved. Now, electrostatic effects (attractive or repulsive) play a more important role with many HILIC columns than is true of RP, acting as a mixed-mode. That means that you have to take salt concentrations and pH into account as well as the concentration of organic solvent when predicting retention. It's more complicated. ERLIC represents an extreme case in which HILIC is performed on a high-capacity ion-exchange column. In this case, the contribution of the charged residues to retention or repulsion is appreciably more significant than is true in regular HILIC, so the situation is even more complicated.

Since so many of the residues in either RP or pure HILIC make some contribution to retention, then the position of any single residue generally is not a major factor in what the retention time is. That's not true in the case of either ion-exchange, ERLIC, or HILIC performed on an ion-exchange column where there is appreciable electrostatic attraction to most of the analytes in the sample (the opposite situation from ERLIC). Since rather few of the residues are charged and hence are involved in the electrostatic effects, their positions in the molecule play a significant role in determining how strongly they can interact with the stationary phase and what the resulting retention times are. That's because many peptides and other solutes migrate through chromatography columns in a highly oriented manner, per this link to a paper on the subject: http://pubs.acs.org/doi/pdf/10.1021/ac100651k
This represents yet another complicating factor that one usually doesn't have to worry about with RP retention time predictions, per my remarks above. Viewed in this light, RP represents the simplest case for which one could develop a predictive algorithm for retention times. Since such algorithms are only picking up speed now, I would imagine that it's going to be a long time before anyone gets a handle on ERLIC and ion-exchange this way.

Andy