Threshold-avoiding proteomics pipeline

We present a new proteomics analysis pipeline focused on maximizing the dynamic range of detected molecules in liquid chromatography-mass spectrometry (LC-MS) data and accurately quantifying low-abundance peaks to identify those with biological relevance. Although there has been much work to improve the quality of data derived from LC-MS instruments, the goal of this study was to extend the dynamic range of analyzed compounds by making full use of the information available within each data set and across multiple related chromatograms in an experiment. Our aim was to distinguish low-abundance signal peaks from noise by noting their coherent behavior across multiple data sets, and central to this is the need to delay the culling of noise peaks until the final peak-matching stage of the pipeline, when peaks from a single sample appear in the context of all others. The application of thresholds that might discard signal peaks early is thereby avoided, hence the name TAPP: threshold-avoiding proteomics pipeline. TAPP focuses on quantitative low-level processing of raw LC-MS data and includes novel preprocessing, peak detection, time alignment, and cluster-based matching. We demonstrate the performance of TAPP on biologically relevant sample data consisting of porcine cerebrospinal fluid spiked over a wide range of concentrations with horse heart cytochrome c.