An integrated LC-MS/MS infrastructure has been conceived and built that emphasizes quality in concert with throughput and turnaround for the high throughput profiling assays that are used to support candidate optimization in drug discovery. Quality, in terms of the breadth of chemical space covered, is achieved via the orchestrated application of tiered multiple ionization modes. The process of checking the fidelity of samples used in the profiling assays at the time and place of use ensures the highest correlation between biological readout and compound structure. The traditional dogma that high-quality quantitative LC-MS requires triple-quadrupole instrumentation (and the associated costs of custom method development for each compound) has been challenged. Ion-trap instrumentation provides near universal collision-induced dissociation (CID) conditions, while the ease of recording product ion spectra via data-dependent MS/MS acquisition provides access to these spectra at no additional cost. The facile access to high-quality product-ion spectra of all compounds under investigation in the profiling assays provides an immediate return, since this approach allows all compounds to be quantitated with the speed, sensitivity, and selectivity that LC-MS/MS affords. Archiving of these spectra into a library (currently > 72,000) that can be searched by both spectra and structure provides a powerful internal tool for impurity and degradant studies. The promise of combining this structure/spectra database with cross-correlation algorithms moves us toward the promise of an empirically based structural-elucidation tool .
The key driver for this automated integrated approach has been the access to processing and instrument control via the COM objects of the Finnigan Xcal-ibur software package. This access to the programming objects of the vendor software has allowed us to focus our programming efforts on those aspects that allow us to meet the specific challenges of the profiling assays by creating new tailored analytical approaches. Without access to these objects we would have had to produce an entire hardware/software solution from the ground up, which would have been impractical. The ability to customize vendor software provides freedom from any vendor preconceptions of how specific analytical challenges should be solved. For example, the development of decision algorithms that extend beyond the simplistic "MW + 1 = compound found" genera of algorithms used for identifying the presence of an expected compound. Recognition that no algorithm can be perfect and an appreciation of the human visual process led us to develop a low false-positive automation approach coupled with single-page color-coded rapid manual review tools.
Was this article helpful?