Two computer programs for developing and improving high-performance liquid chromatographic methods, DryLab G and LCSIM, have recently been described. The accuracies of these two programs were examined using experimental (o-phthalaldehyde-derivatized amino acids) and synthetic data. DryLab G, which uses gradient data for input, correctly predicted retention times for various gradient and isocratic separations. Predicted retention times for the simulation of certain isocratic conditions are susceptible to errors in the measured dwell volume, but the prediction of resolution is not seriously affected. LCSIM uses isocratic data for input, and predicted gradient retention times are affected by the accuracy of the measured dwell volume. The resolution of closely eluting analytes was usually predicted within a small fraction of the peak width, i.e., with negligible errors.

The separation of seven o-phthalaldehyde-derivatized amino acids by reversed-phase gradient elution was studied as a function of gradient time and mobile phase flow-rate. The resulting separations were compared with those from computer simulation (Drylab G). Predictions of retention time via computer simulation were found to be quite accurate, being about ±0.7% for retention time and about ±7% for retention-time differences (resolution). Predictions of band width were accurate within about ±15% for all but the steepest gradients (b > 1.0). Consequently, the ability of computer simulation to predict chromatograms reliably as a function of gradient conditions and flow-rate was confirmed for a sample that is representative of “real life”. For very steep gradients (b > 1.0), significant errors in band width were observed. The source of these errors could arise from various effects which are discussed.

Using DryLab G to identify a "hidden" band in a protein separation. Complex chromatograms that result from reversed-pahse gradient elution often exhibit changes in band order when the gradient steepness is changed. This complicates the interpretation of the resulting separation, and prevents the application of computer simulation for method development. A simple procedure based on normalized band areas was used to match bands between runs where the gradient steepness has been changed. In one example involving a Thermus aquaticus ribosomal protein sample, it was possible to find an additional band that was not apparent in tw initial experimental runs with different gradient slopes.

The herbicide fluroxypyr 1-methylheptyl ester was separated from its acid form (fluroxypyr) and two soil metabolites (a pyridinol and a methoxypyridine) by high-performance liquid chromatography (HPLC) with the acid of Drylab G. A difference in retention time of at least 6 min between each compound was achieved, to allow complete separation of radiolabeled components during collection of 1-min fractions. The total elution time was 31 min. Actual HPLC retention times differed from DryLab G predictions by 0.5 min or less.

High-performance liquid chromatographic (HPLC) methods for analyzing new drugs and their synthetic intermediates are needed as the synthesis is optimized and scaled up from making milligram amounts for initial evaluation of biological activity to producing kilogram amounts of the drug for thorough testing purposes. The most efficient solution is a single HPLC method that can be used for each step of the synthesis. A practical approach for the development of a single HPLC method is the use of computer-assisted method development to maximize the resolution within a reasonable analysis time. The computer program DryLab I was used in the development of an HPLC assay for the synthetic intermediates of a leukotriene inhibitor. The use of DryLab I with binary mixtures of organic solvents in the organic portion of reversed-phase HPLC systems is reported. With the retention data from two initial analyses, resolution can be optimized as a function of solvent strength.

Using DryLab G to identify a "hidden" band in a protein separation. Complex chromatograms that result from reversed-pahse gradient elution often exhibit changes in band order when the gradient steepness is changed. This complicates the interpretation of the resulting separation, and prevents the application of computer simulation for method development. A simple procedure based on normalized band areas was used to match bands between runs where the gradient steepness has been changed. In one example involving a Thermus aquaticus ribosomal protein sample, it was possible to find an additional band that was not apparent in tw initial experimental runs with different gradient slopes.