Liquid chromatography-mass spectrometry (LC-MS) is a widely used technique in the field of proteomics for biomarker discovery and protein identification. An LC-MS experiment determines the components of a chemical mixture and the result can be represented as a two-dimensional image, where the LC retention time and the mass-to-charge ratio are the abscissa and the ordinate respectively, while the signal intensities indicate the abundance of detected chemicals. In many LC-MS based applications, it is often required to compare samples from different experiments, and thus multiple LC-MS images. Since there is always a non-linear LC retention time shifting across LC-MS images, alignment between images is needed prior to performing any analysis.

In this thesis, we introduce a multi-...[ Read more ]

Liquid chromatography-mass spectrometry (LC-MS) is a widely used technique in the field of proteomics for biomarker discovery and protein identification. An LC-MS experiment determines the components of a chemical mixture and the result can be represented as a two-dimensional image, where the LC retention time and the mass-to-charge ratio are the abscissa and the ordinate respectively, while the signal intensities indicate the abundance of detected chemicals. In many LC-MS based applications, it is often required to compare samples from different experiments, and thus multiple LC-MS images. Since there is always a non-linear LC retention time shifting across LC-MS images, alignment between images is needed prior to performing any analysis.

In this thesis, we introduce a multi-resolution LC-MS image alignment scheme for synchronizing LC-MS images. The multi-resolution scheme employs one of two popular time alignment algorithms, Dynamic Time Warping (DTW) or Correlation Optimized Warping (COW), as the optimization algorithm, while Kullback-Leibler Distance (KLD) is used as the local similarity measure. We have validated the proposed scheme using two real-world data sets, which include both high mass accuracy and low mass accuracy LC-MS images. In addition, we have also compared the results with an existing alignment algorithm, the Dynamic Time Warping-Component Detection Algorithm (DTW-CODA). The experiments show that the proposed scheme outperforms DTW-CODA, and is a promising approach for aligning both high mass accuracy and low mass accuracy LC-MS images.