Enhancing the Data, Information and Knowledge from Proteomics Experiments through Biomedical Informatics
A systems approach to understanding disease begins with data that contains information from which knowledge is developed and ultimately provides wisdom and understanding to human health and disease. Systems biology is a fast growing area of research to promote translational medicine and is discussed as an alternative to reductionism. In this work, shotgun proteomics is utilized along with systems thinking to quantify numerous proteins simultaneously and presents an unbiased, collective view of human health and disease. Following the capture of large amounts of proteomics data as mass spectra and assignment of peptide sequence, a novel strategy for ratio determination based on power spectral filtering and Levenberg-Marquardt fitting has been developed and is discussed here. This strategy is then applied here for several studies investigating a variety of topics, including: traumatic brain injury, Alzheimer’s disease and diabetic nephropathy and the discovered knowledge and understanding is discussed. Also, parametric and non-parametric statistical methods for assigning significance are applied here to determine the key proteomic changes between proteomes and enhance knowledge discovery. Further, an empirical Bayesian method of combining multiple proteomics experiments as an application of meta-analysis is used to reduce the biological variability between experiments in a rodent model of diabetic kidney disease in type 1 diabetes. After determining the significant changes, the data are converted to information prior to comparing differences in order to enhance knowledge discovery and facilitate trans-omics research. Vector space models are developed and applied here to allow comparisons based on information as opposed to data and removes inherent difficulty in comparing systems biology studies. This work has provided substantial enhancement to proteomics and systems biology methods and applications.