Composite Peak Scoring for Improved Charged Aerosol Detector-Based Universal Quantification of Drug Libraries

To accelerate drug discovery, scientists must innovate within the design-make-test-analyze (DMTA) loop. Robotically synthesized drug libraries are an example acceleration technique. Molecules from these libraries can be directly transferred to biological assays, purified then assayed, quantified then assayed, or combinations thereof. At Amgen, thousands of biologically active molecules have been produced via library synthesis and subjected to our mass-directed microgram scale high-throughput purification process (μgHTPP). After purification, reaction yield and purification recovery remain unknown, so biological activity cannot be directly determined. To address this need, a charged aerosol detector (CAD) is used as a "universal detector" during postpurification quality control to quantify the product in each well. The CAD is touted as having excellent universality for nonvolatile analytes based on flow injection analysis. However, all real-world separations require a high-resolution separation step prior to analysis to isolate the product. Until this report, the effect that chromatography has on the CAD's output was largely ignored. These effects are shown to change the peak area by greater than 3-fold for the same sample, violating proportionality between the peak area and analyte concentration. We show that the peak area and peak height carry unique information for CAD-based universal quantitation. A composite Z-score is introduced, exploiting information from both metrics. Calibration with Z-score reduced error ranges from 0.13-55% (area) and 0.07-54% (height) to 0.03-28%, eliminating all error >30% across a 195-point validation set. Results from 335 compounds demonstrate Z-scoring's strong impact on quantitation and activity values.