The development and optimization of liquid chromatography (LC) separations can be time-consuming and costly, often requiring several steps, including literature research, column selection, method scouting, method development, and method optimization. In an effort to eliminate these steps, an instrument-free, software modeling tool that gives users the ability to select compounds from a database and instantly model a separation on different column stationary phases was developed. Optimization of the model can be performed while maintaining critical pair separations by adjusting for instrument/system effects (e.g., dwell volume and extra column volume); mobile phase preferences; number of gradient steps; and more. The initial database consists of a Drugs of Abuse (DoA) library containing approximately 250 compounds with plans to continually expand the utility.
To build the chromatogram modeler, a DoA library containing approximately 250 compounds was created. Data was collected using a standard HPLC system coupled to a triple quadruple mass spectrometer (LC-MS/MS). Retention times were first collected using a fast/slow gradient; 30°C/60°C column temperature points; and ACN/MeOH mobile phases on a single column dimension (50 x 2.1 mm, 2.7 µm). Some additional data points outside of these runs were also collected for the development of a semiempirical correction factor that was used to improve modeling accuracy. To assess the accuracy of the modeler, experiments comparing compound retention time values between wet-lab and modeled data were conducted. After the initial DoA library was built, the modeler was evaluated over four increasingly more complex stages of verification. In the final, most complex stage, new compounds not previously part of the initial DoA library were added and then compared by testing two different column dimensions, two different columns lengths, two different mobile phases, two different stationary phases, three different gradients programs, and three different temperatures against modeled retention time values. Because the semi-empirical correction factor was developed using only the original library compounds, this stage assessed the viability of adding future compounds to existing libraries.
During software development, the acceptance criteria for retention time agreement between wet-lab and modeled values was set at +/- 15 seconds. This range was chosen because it represents a typical MRM window. In the most complex portion of the verification, 704 retention time data points were collected in total for the 25 compounds used in the evaluation. Only 13 data points exceeded the +/- 15 second window with no compounds missing acceptance criteria by more than five seconds, giving an overall pass rate of 98.2%.
An online chromatographic modeling tool was successfully developed that allows users to select columns and compounds for separation. A modeled chromatogram and instrument-ready conditions are automatically generated and can be further optimized by users.
