Prior Knowledge for Discovery Proteomics Data Interpretation

Discovery proteomics usually refers to the search for potential biomarkers of disease from clinical samples such as blood serum, plasma, urine or cerebrospinal fluid (CSF). These samples are typically analyzed using high-throughput Mass Spectral (MS) technologies yielding highdimensional data that needs interpretation in order to identify a small number of protein/peptide biomarkers predictive of disease states. This research provides an analysis of a discovery process leading to identification of potential biomarkers of a neurodegenerative disease called Amyotrophic Lateral Sclerosis (ALS). Task-specific domain knowledge is identified that can be used as priors to guide the search. Ongoing research involves the identification of such prior knowledge, and its use within rule learning and probabilistic machine learning systems to enhance the accuracy of biomarker panels predictive of disease states.