New statistical algorithms for clinical proteomics

Abstract

Background: Mass spectrometry based screening methods have
been recently introduced into clinical proteomics. This boosts the development
of a new approach for early disease detection: proteomic pattern analysis.

Aim: Find, analyze and compare proteomic patterns in groups
of patients having different properties such as disease status or
epidemio-logical parameters (e.g. sex, age) with a new pipeline to enhance
sensitivity and specificity.

Problems: Mass data acquired from high-throughput platforms
frequently are blurred and noisy. This extremely complicates the reliable
identification of peaks in general and very small peaks below noise-level in
particular.

Approach: Apply sophisticated signal preprocessing steps
followed by statistical analyzes to purge the raw data and enable the detection
of real signals while maintaining information for tracebacks.
Results: A new analysis pipeline has been developed capable
of finding and analyzing peak patterns discriminating different groups of
patients (e.g. male/female, cancer/healthy). First steps towards distributed
computing approaches have been incorporated in the design.