Mastering data pre-processing for accurate quantitative molecular profiling with liquid chromatography coupled to mass spectrometry

Identification of perturbations in biological systems is the corner stone of biomedical research. Measuring changes at molecular level has allowed biomedical research to understand key physiological and molecular mechanisms in living systems. Proteomics allows analysis of these system level molecular changes and interactions in complex biological samples. Data dependent acquisition (DDA) is the most widely used approach for comprehensive profiling of proteins and metabolites in bottom-up proteomics experiments acquired using LC-MS/MS. This thesis describes novel quality assessment methods used to study orthogonality in the retention time domain (separation dimension) and vital pre-analytical factors affecting the ion intensity (readout dimension) of LC-MS/MS datasets. Thus following statements form the main goals of the thesis -
- Summary of various data pre-processing steps involved in the treatment of label-free LC-MS(/MS) datasets, obtained for a typical proteomics experiment.
- Describe the MS1 stage of a LC-MS(/MS) dataset as a second order tensor (three dimensional data). Discuss various physio-chemical origins and effects of
orthogonality in the two separation dimensions (m/z and retention time) and readout
dimension (ion intensity).
- Presentation of a quality assessment approach, which evaluates orthogonality in the retention time dimension for a pair LC-MS(/MS) chromatograms following correction
of monotonic shifts.
- Proposition of a method to annotate unmatched spectra based on the concept of “identification transfer” after correction of monotonic shifts between datasets and
assess the FDR associated in matching features based on retention time and m/z coordinates between datasets.
- Application of Anova-Simultaneous Component Analysis (ASCA) to determine, which pre-analytical factors influences the ion intensity domain of LC-MS features.