Technology Advances
Developers of new methods for liquid chromatography battle a range of variables: buffer concentration, flow rate, reagents, and more. To figure out how to handle each variable, Margaret Antler, technical marketing specialist for mass spectrometry and chromatography at Advanced Chemistry Development (ACD/Labs) in Toronto, Ontario, Canada, says, “People used to guess.” She adds, “If you’ve been developing LC methods for a long time and know the chemical structure of what you want to get from an LC run, you might look at that structure and then have a feel for, say, the optimal pH for a separation method.” But guessing is not the best or fastest way to optimize an LC method (See “Academic Insights”).
Some methods, for example, might require balancing a separation gradient and temperature. “If you just optimize one of these variables at a time,” says Antler, “you won’t necessarily get the best combination. But you can if you model it.”
To give developers a more objective approach, ACD/Labs created its AutoChrom software, which is currently shipping as a beta version. Using a structured approach to method screening, and mathematical models and predictions for further optimizations—based on the chemical structure of the analytes, knowledge of the column’s characteristics, and linear optimization models—this software predicts the parameters that will create improved LC methods. As Antler says, “Our software and predictions let you develop better methods with less effort.”
Nonetheless, AutoChrom is not automatic method development. It can model an experiment, and then suggest the next step, but control of the experiment remains with the chromatographer. For example, the user might disagree with AutoChrom’s suggestion, and then the software can try out the user’s idea.
The in silico machinery behind AutoChrom also provides other perks. For example, it can bring in LC-MS data to track the peaks from one experiment to the next. “This is a new idea for most chromatographers,” says Antler. “MS-based peak tracking can help you find all components in your sample, and pick the best column and mobile phase combination after screening experiments.” For example, many AutoChrom users work in the pharmaceutical industry, and they use this software to develop stability-indicating methods for new drugs.
Getting up to speed on all of AutoChrom’s features, though, can take some time. “New users can be up and running quickly,” says Antler, “but it takes some time to learn all of the features.” So, ACD/Labs provides training—which ranges from online consultations to on-site classes.
Overall, AutoChrom can change LC method development from guessing to guided.
About the Author
May is a publishing consultant for science and technology based in Minnesota.
This article was published in Drug Discovery & Development magazine: Vol. 10, No. 9, September, 2008, pp. 14.
Academic Insights
Modeling adds efficiency and objectivity to developing chromatographic methods. “You change the process development scheme from a trial and error plus experience–based scheme to a model-assisted, process development scheme, where the experiments you do are experiments to determine model parameters and then you use a simulator to determine the optimal process conditions,” says Jørgen Mollerup, docent, department of chemical and biochemical engineering, Technical University of Denmark.
As an example of such a process, Richard Brereton, PhD, professor in the school of chemistry at the University of Bristol, UK, says, “One can try to connect physical properties of columns to separations, such as retention time and peak widths of compounds. One can also introduce parameters such as pH, mobile phase composition, etc. This means that the separation efficiency of columns for specific classes of compounds can be predicted, reducing the need for extensive experimentation.”
Despite such benefits of modeling, Florian Dismer, a graduate student working on characterizing adsorption in ion exchange chromatography at the University of Karlsruhe in Germany, says, “We are currently still far away from designing processes on a computer from scratch.” But he adds, “in silico methods can help in the future to minimize experimental efforts needed for optimizing downstream processing.”
|
|
