Background: A major challenge in high-throughput clinical and toxicology laboratories is the reliable processing of chromatographic data. In particular, the identification, location, and quantification of analyte peaks needs to be accomplished with minimal human supervision. Data processing should have a large degree of self-optimization to reduce or eliminate the need for manual adjustment of processing parameters. Ultimately, the algorithms should be able to provide a simple quality metric to the batch reviewer concerning confidence about analyte peak parameters.
Content: In this review we cover the basic conceptual and mathematical underpinnings of peak detection necessary to understand published algorithms suitable for a high-throughput environment. We do not discuss every approach appearing in the literature. Instead, we focus on the most common approaches, with sufficient detail that the reader will be able to understand alternative methods better suited to their own laboratory environment. In particular it will emphasize robust algorithms that perform well in the presence of substantial noise and nonlinear baselines.
Summary: The advent of fast computers with 64-bit architecture and powerful, free statistical software has made practical the use of advanced numeric methods. Proper choice of modern data processing methodology also facilitates development of algorithms that can provide users with sufficient information to support QC strategies including review by exception.
© 2015 American Association for Clinical Chemistry.