Paper detail

Bayesian hierarchical reconstruction of protein profiles including a digestion model

Introduction : Mass spectrometry approaches are very attractive to detect protein panels in a sensitive and high speed way. MS can be coupled to many proteomic separation techniques. However, controlling technological variability on these analytical chains is a critical point. Adequate information processing is mandatory for data analysis to take into account the complexity of the analysed mixture, to improve the measurement reliability and to make the technology user friendly. Therefore we develop a hierarchical parametric probabilistic model of the LC-MS analytical chain including the technological variability. We introduce a Bayesian reconstruction methodology to recover the protein biomarkers content in a robust way. We will focus on the digestion step since it brings a major contribution to technological variability. Method : In this communication, we introduce a hierarchical model of the LC-MS analytical chain. Such a chain is a cascade of molecular events depicted by a graph structure, each node being associated to a molecular state such as protein, peptide and ion and each branch to a molecular processing such as digestion, ionisation and LC-MS separation. This molecular graph defines a hierarchical mixture model. We extend the Bayesian statistical framework we have introduced previously [1] to this hierarchical description. As an example, we will consider the digestion step. We describe the digestion process on a pair of peptides within the targeted protein as a Bernoulli random process associated with a cleavage probability controlled by the digestion kinetic law.