Analysis of sets and collections of Peptide Mass Fingerprint data

Abstract

Recent advances in genomics, which outstanding achievements were exemplified by the complete sequencing of the human genome provided the infrastructure and information enabling the development of several proteomic technologies. Currently no single proteomic analysis strategy can sufficiently address the question of how the proteome is organised in terms of numerical complexity and complexity generated by the protein-protein interactions forming supramolecular complexes within the cell. In order to bring a detailed structural/functional picture of these complexes in whole genomes, cells, organelles or in normal and pathological states several proteomic strategies can be utilised. Combination of technologies will bring a more detailed answer to what are the components of certain cellular pathways (e.g.: targets of kinases/phosphatases, cytoskeletal proteins, signalling molecules), how do they interconnect, how are they modified in the cell and what are the roles of several complex components in normal and disease conditions. These types of studies depend on fast and high throughput methods of protein identification. One of the most common methods of analysis is mass spectrometric technique called peptide mapping. Peptide mapping is the comparison of mass spectrometrically determined peptide masses of a sequence specific digest of a single protein or peptide of interest with peptide masses predicted from genomic databases. In this work several contributions to the computational analysis of mass spectrometric data are presented. During the course of my studies I looked at the distribution of peptide masses in sequence specific protein sequence digests and developed a simple mathematical model dealing with peptide mass cluster centre location. I have introduced and studied the methods of calibration of mass spectrometric peak-list without resorting to internal or external calibration samples. Of importance is also contribution of this work to the calibration of data produced in high throughput experiments. In addition, I studied how filtering of non-peptide peaks influences the identification rates in mass spectrometric instruments. Furthermore, I focused my studies on measures of spectra similarity which can be used to acquire supplementary information, increasing the sensitivity and specificity of database searches.